Ad hoc network is regarded as an optimal solution to establish network on poor quality environments such as stricken areas due to its simple and low cost operation. However, conventional routing methods cause centralization of packet transmission load except for the occurrence of dynamic network topology changes. In this paper, we propose a novel location-based load balancing method that can decentralize packet transmission load of each node in the whole networks. The proposed method opportunistically selects a route for data transmission from several route candidates prepared for load balancing. The performance and the efficiency of the proposed method are also clarified by computer simulation.

In ad hoc networks, frequent packet losses and topological changes cause unstable communications. To solve this problem, OR (opportunistic routing) has been proposed. In OR, each terminal decides whether to forward a received packet based on a random backoff time. However, forwarder terminals must wait for the expiration of the random backoff timer before forwarding the packet. Moreover, it is difficult to obtain a transmission path diversity under a sparse environment. This paper proposes a forwarder selection method that uses the prioritized forwarders to reduce the backoff time. In addition, the proposed method also performs a retransmission control which is fit for the sparse environment. The performance and the effectiveness of the proposed method are clarified by computer simulation.

In a typical network service, the users request network resources so that their utility is maximized. On the other hand, the network operator assigns network resources to users without expecting the cooperation among users. This is a scramble for limited network resources, and it yields "tragedy of commons". Thus, the efficient allocation of network resources is a difficult problem. In this paper, we regard the resource allocation among users in the wireless network as a social dilemma, and propose a solving method based on the framework of mobility management.

To increase the availability of networks, there are a number of studies focusing on ensuring stable network environments on the unexpected occurrence of huge disaster. Above all, ad hoc networks are well suited for these situations due to its simple and rapid deployment of wireless networks at low costs. Several location-based routing protocols are proposed to enhance the performance by establishing stable transmission route. Scalability, however, are not taken into consideration due to the limitation of node mobility management on most location-based studies. In this paper, we propose a novel location-based routing method can improve communication efficiency by performing opportunistic link selections. Performance of the proposed method is verified by computer simulation.

An ad hoc network is characterized as a self-organized multi-hop wireless network. It does not rely on any pre-existing network infrastructures. Instead of this, each node can participate in the routing of packets to other nodes. However, each node in ad hoc networks is typically powered by batteries with a limited energy supply, and it is difficult to replace or to recharge them. Thus, it is crucial to consider how to conserve energy to maximize the network lifetime as an issue in the design of routing protocols. In conventional routing protocols, however, shortest-route algorithms are commonly used. Without considering energy consumption, nodes that are overused would drain their energy and the whole network would become partitioned. As concerns this problem, many battery-aware routing protocols have been proposed from a variety of perspectives. In this paper, we propose a novel enhanced max-min residual battery-aware routing protocol called EMAODV. It is expected to prolong the network lifetime by selecting an optimal route based on the battery conditions of the whole network. By EMAODV, the routes with insufficient battery energy are always avoided. If all the nodes in ad hoc networks have sufficient battery energy, a shortest route is selected. From the results of the performance evaluation, EMAODV shows a better performance in terms of prolonging the network lifetime.

Dropping packets is one of the most devastating threatens on a performance of ad hoc networks. Among various motivations of dropping packets, selfishness and malicious attack are the main two factors. However, there are still some other cases of dropping packets caused by networking context such as mobility, congestion, and traffic load. Incentive mechanism is a typical method to induce node to participate in relaying packets for other adjacent nodes. However, most of them neglect a noise of dropping packets by networking context or just omit this important factor for brevity. In this paper, a principal-agent theory based game theoretic incentive method is proposed for encouraging each node contributing into packet forwarding, which will result in a high performance cooperative network. Holmstrom-Milgrom model, which is a simplified principal-agent model with a linear reward scheme, is implemented to model the relationship between two adjacent nodes. In each two pair-wise nodes model, the node requests for relaying packets is seen as a principal node while the node forwarding packets is denoted as an agent node. A credit given by principal node to agent node as incentive compensation is the contract, which is designed by principal node based on agent node's effort in forwarding packets. Dropping packets by networking is also taken into consideration as an exogenous uncertainty. It is proven that there is an optimal incentive strategy can encourage packet forwarding even under noise effect. Numerical evaluations are also conducted to evaluate the performance of the proposed method.

Due to the inherent broadcast nature of wireless communication, ad hoc networks are well suited for multicast that can more efficiently deliver information from source nodes to many client nodes simultaneously. In ad hoc networks, however, links are established and disconnected dynamically due to the transmission interference and node movement. In conventional multicast routing protocols, the topology dynamics cause route reconstruction, and it may lead delay increase and delivery ratio degradation. In this paper, a novel multicast routing protocol is proposed. This protocol can improve packet delivery rate of ODMRP(On-demand multicast routing protocol) by making use of link failure detection mechanism as well as nodes' mobility prediction based local repair mechanism. The performance and the efficiency of the proposed method are also clarified by computer simulation.

Most image recommendation services in SNS (Social Network Services) use collaborative filtering to calculate the similarity of contents. Collaborative filtering requires high computing power. It is difficult to calculate the recommendation indices immediately. In this paper, a new recommendation method is evaluated. It uses the genre information of the image content. The Web site application software is implemented to evaluate the accuracy of the recommendation.

In ad hoc networkserminals' mobility may cause dynamic change of network topology, which makes it difficult to establish stable transmission routes. As concerns this problem, opportunistic routing (OR), which does not rely on any specific route and forwards packets opportunistically using stable links, has been proposed. In order to achieve efficient packet forwarding in ad hoc networks, this paper proposes a new OR protocol based on the estimated end-to-end delay that is calculated from terminals' queue lengths and links' transmission success rates. The proposed method tries to use the terminals that have the shorter delay to the destination terminal as forwarders for realizing shorter end-to-end delay and higher throughput. The performance and the efficiency of the proposed method are also clarified by computer simulation.

By the rapid spread of smartphones, huge traffic data occurs in 3G networks. Traffic load is serious and becomes a big problem. A lot of researches have been done for the overload of the traffic. One of the approaches is the traffic control by using the user's bebaviour. If a user does not satisfy the network quality, he wants to move to the best access point to improve his utility. User guidance service navigates the best access point. Two incentive control methods are shown in this paper. The model of the user's moving behaviour is proposed. From the questionnaire, the relation between QoS and willingness to move is shown.

IPTV services are becoming popular. However, the pricing methods for IPTV services have seldom been discussed. In our previous studies, we proposed the pricing methods for maximizing the social welfare and for maximizing the provider's profit in VOD-type IP broadcasts services. In this paper, the IPTV service models that defined by IPTV Form in Japan are shown. The viewing behaviours are modelled for each IPTV service model. The pricing methods for maximizing the user utility are discussed.

Mobile ad hoc network (MANET) is a self-organized, infrastructure less, multi-hop network, which is highly vulnerable to various attacks. Among various attacks, DoS (Denial of Service) attack that may lead to the collapse of the network performs as a disaster in the network. Wormhole attack, a particularly devastating DoS attack, is taken by two or more colluding nodes that create a virtual tunnel to do further malicious actions. The tunnel emulates shorter links in the network and can be employed to transport packets between end points of the tunnel. By using wormhole attack, malicious nodes can not only eavesdrop packet contents but also modify or drop legitimate packets. Wormhole attack is even more harmful than other kinds of DoS attacks for it can be launched even without getting any cryptographic keys or compromising the existing nodes. This paper proposes a mobile agent based strategy to overcome not only the closed wormhole attack but also the half open and open wormhole attack. Both wormhole using encapsulation and wormhole using out-of-band channel are taken into consideration and the proposed method shows a good performance in detecting the malicious nodes.

DDoS attack is becoming a serious problem in mobile ad hoc networks. In the attack, widely distributed malicious nodes flood packets into the network, and exhaust resources of the network such as bandwidth, computing power, etc. As concerns this, some defence methods such as confirmation-based, traffic analysis-based methods are proposed to detect the malicious nodes. However, the conventional defence methods may cause delay increase and nodes' load increase. In order to solve these problems, this paper proposes a novel defence method against the DDoS attack which detects malicious nodes using relay priority based on time interval of route requests. The performance and the effectiveness of the proposed method are shown by computer simulation.

In ad hoc networks, terminals' mobility may cause dynamic changes of network topology and terminal density which may increase the traffic at a certain terminal or in a certain location. As concerns this problem, some routing-based load balancing methods which are based on traffic or transmission success rate have been proposed. However, these conventional methods consider only the load on the transmission route, and this may increase the load at neighbouring terminals of the route. In order to achieve efficient load balanced routing in ad hoc networks, this paper proposes a load balancing method based on an estimated traffic around each terminal. The proposed method monitors the neighbouring terminals' transmission and estimates a relative load based on it for better route selection. The performance and the efficiency of the proposed method are also clarified by computer simulation.

In ad hoc networks, dynamic changes of the network topology caused by the terminals' mobility may increase the traffic load on a certain terminal or in a certain location. As concerns this problem, some traffic-based routing mechanisms to balance the load have been proposed. However, it is hard to adapt them to the environment where the traffic changes in a short period. Furthermore, these conventional methods do not consider the load on the neighbouring terminals of the transmission route. In order to achieve efficient load balancing in ad hoc networks, this paper proposes a load balancing method based on neighbouring terminals' traffic. The proposed method controls the transmission rate depending on the neighbouring terminals' load and the transmission route. The performance and the efficiency of the proposed method is also clarified by computer simulation.

Tolerable waiting time for content download depends on content size and situations. It is because of the difference of user's utility. If EDF (Earliest Deadline First) is used instead of FIFO, users' utility will increase. However, it needs much processing. In order to avoid it, QEDF (Quasi Earliest Deadline First) is considered in this paper. It does not need much processing. Assuming that the portable devices are used for content download, users' utility of QEDF is almost the same as that of EDF.

Wormhole attack is becoming a serious problem in ad hoc networks. In the attack, a pair of malicious nodes establishes a tunnel between two end points, and collects the transmission route of the neighbouring nodes. Therefore, the data on the route might be wiretapped, falsified, and annulled by the malicious nodes. To avoid the attack, some defence method is necessary to detect the malicious nodes. However, the conventional defence methods may cause delay increase, and their scalability is limited. In order to solve these problems, this paper proposes a novel defence method against the wormhole attack which detects neighbouring malicious nodes using observation report by normal nodes. The performance and the effectiveness of the proposed method are shown by computer simulation.

In ad hoc networks, it is hard to realize the centralized management of network state, and dynamic changes of the network topology caused by the terminals' mobility may increase the traffic load on a certain terminal or in a certain location. As concerns this problem, some routing mechanisms to balance the network load have been proposed. However, it is hard to adapt them to the environment where the traffic changes in a short period. Furthermore, these conventional methods only consider the load on the transmission routes. For achieving efficient load balancing in ad hoc networks, this paper proposes a load balancing method based on the activity of neighbouring terminals. The proposed method controls the transmission rate depending on the states of the neighbouring terminals and of the transmission route. The performance and efficiency of the proposed method is also clarified by computer simulation.

The quality of network is actually ultimately evaluated by users who use the service supplied by it. Although the conception of QoE (Quality of Experience) has been proposed and considered important recently, it has not commonly used to evaluate the quality of network yet. QoE usually has a nonlinear relationship with QoS (Quality of Service) which is usually a set of objective indices for evaluation. In this paper, how the optimization of OSPF routing reflects in both QoS and QoE is discussed by using the example of content download, and it is shown that there is an improvement of 60% in QoS and 10% in QoE through the optimization of OSPF routing. From the viewpoint of users' utility, the complete optimization of OSPF routing is not necessary, and the quasi-optimization is enough.

In ad hoc networks, it is hard to unify the management of network state due to an autonomously decentralized manner for network construction mechanism by mobile terminals. Topological changes may increase the traffic load on a certain terminal or location, and thus that will degrade the fairness of the whole network. As concerns this problem, the methods that modify routing mechanisms to balance the traffic load have been proposed. However, they need extra control sequence for the route modification and hardly adapt to the environment where the traffic changes in a short period. In this paper, to balance the load efficiently in ad hoc networks, cross-layer transmission rate control method is proposed. The proposed method controls the transmission rate, by limiting the maximum, depending on the state of the transmission route, and thus avoids the concentrated traffic and collisions. The performance and efficiency of the proposed method is also clarified by computer simulation.

The network service employs multiple transmission media to convey information to the users. The quality of experience (QoE) should be considered among different media that transfer service content. QoE of telecommunication services has been previously discussed by QoS mapping. While QoS is the ability to guarantee a certain level of performance to a data flow, QoE presents the degree of satisfaction of the user and can be often expressed by subjective evaluation such as the mean opinion score (MOS). From the viewpoint of network provider, it is possible to predict the user's bihaviour by QoE function. In this paper, we introduce the previous research of QoE measuring, show the priority control of contents delivery for the maximizing users' utility.

The network service employs multiple transmission media to convey information to the users. The quality of experience (QoE) should be considered among different media that transfer service content. QoE of telecommunication services has been previously discussed by QoS mapping. While QoS is the ability to guarantee a certain level of performance to a data flow, QoE presents the degree of satisfaction of the user and can be often expressed by subjective evaluation such as the mean opinion score (MOS). From the viewpoint of network provider, it is possible to predict the user's bihaviour by QoE function. In this paper, we introduce the previous research of QoE measuring, show the priority control of contents delivery for the maximizing users' utility.

In VOD type IPTV services, if the number of users' requests is more than the server capacity (maximum number of streams), some users' requests will be blocked. If the viewing price is usage-based, the provider will lose the revenue as the blocking probability increases. The server capacity should be increased in order to decrease the blocking probability. However, if the cost to increase the server capacity is high, the viewing price will be high. Then, the number of users' requests will decrease, and the provider's revenue will decrease. Therefore, it is important to set the viewing price by considering both the server cost and the blocking probability. This paper analyzes this relation and shows the viewing price and the server capacity that maximize the provider's profit.

In ad hoc networks, since each terminal transmits data using radio devices, segment loss frequently occurs owing to radio interference or node mobility. TCP Reno uses a segment loss to detect congestion and tries to avoid it by shrinking the congestion window size. In ad hoc networks, however, segment losses are caused by not only congestion but also interference and node mobility. Thus, unnecessary congestion avoidance decreases communication efficiency. As concern this, to improve communication efficiency, authors proposed RTT-based window control method. However, since a parameters for window control is a constant in the method, there is a possibility that the threshold would not be appropriate value for some cases. In this paper, the effect of the parameters on communication efficiency is analyzed to set the optimal parameters. This will yield better performance.

IPTV services are becoming popular in the Internet. However, the optimal pricing method for IPTV has seldom been discussed. In VOD type IPTV services, the maximum number of streams depends on the server capacity. If the number of users' requests is more than the maximum number of streams, users' requests will be blocked. If the number of subscribers increases, the provider's revenue will increase. However, the blocking probability will also increase, and the user's utility will decrease. In order to maintain QoS, the provider must have more server capacity. This means that the server cost will increase. This paper shows the relation between the viewing price and the users' benefit, and that between the viewing price and the provider's profit for various server capacity. The optimum viewing price for the users' benefit and that for the provider's profit are different. To solve this problem, Nash bargaining solution is calculated by using game theory.

Recently, the image content such as movie, news, etc. is distributed to portable telecommunication devices. In this case, the relation between image quality and utility, and that between image quality and waiting time allowance will depend on the viewing purpose. In this paper, such relation is analyzed by using subjective evaluation and questionnaire survey.

In a content download service, it is effective to perform both a high-speed delivery and a low-speed delivery simultaneously by multiple servers. However, in this distribution method, it is necessary to set a long delivery interval for all servers. In the previous research, it is shown that the optimal delivery interval depends on the delivery speed, and it is ideal to set a short delivery interval to a high-speed server. For this reason, the problem of this delivery method is that useless standby time occurs on the high-speed server. In this paper, a delivery method to solve this problem is suggested. Servers divide content into content blocks, and give a block number to each content block. Each delivery servers alternate the content block order between ascending order and descending order. The results prove that the proposed method can increase users' utility compared with conventional methods.

IP broadcast services are becoming popular in the Internet. However, the optimal pricing method for IP broadcasting is not discussed. This paper analyses flat-rate price of VOD type IP broadcast services, and examines the point of agreement between provider's revenue and user's benefit by using the negotiation problem of game theory. This analysis pays attention to the request blocking rate as the quality of service. The negotiation model between user and provider is defined, and the points of agreement for various request blocking rates are obtained by the agent simulator. As a result, the optimal prices are shown.

IP broadcast services are becoming popular in the Internet. However, the optimal pricing method for IP broadcasting is not discussed. This paper analyses flat-rate price of VOD type IP broadcast services, and examines the point of agreement between provider's revenue and user's benefit by using the negotiation problem of game theory. This analysis pays attention to the request blocking rate as the quality of service. The negotiation model between user and provider is defined, and the points of agreement for various request blocking rates are obtained by the agent simulator. As a result, the optimal prices are shown.

IP broadcast services are becoming popular in the Internet. However, the optimal pricing method for IP broadcasting is not discussed. This paper analyses flat-rate price of VOD type IP broadcast services, and examines the point of agreement between provider's revenue and user's benefit by using the negotiation problem of game theory. This analysis pays attention to the request blocking rate as the quality of service. The negotiation model between user and provider is defined, and the points of agreement for various request blocking rates are obtained by the agent simulator. As a result, the optimal prices are shown.

Packet contention occurs at a switching node in optical packet switching networks. By using fibre delay line (FDL) buffers, packet contention can be avoided. In case of synchronous networks, if the length of FDLs is properly selected according to the fixed length of packets, no waste of FDL use will occur. FDL will be used efficiently to avoid packet contention. However, if the current Internet packets which have variable lengths are fragmented and embedded into the fixed length optical packet of synchronous networks, it is considered inefficient due to the waste within the last optical packet where the last fragmented payload is always shorter than the fixed optical packet size. This will degrade the performance. In this paper, through simulations, we quantitatively evaluate the performance between fixed length packets and variable length packets in optical packet switching. Our motivation in this paper is to observe the possible clues to improve the FDLs efficiency and system performance in both asynchronous and synchronous optical packet switching networks.

In ad hoc networks, there is a concept of reward given to relay nodes to compensate for their loss of power. When implemented as a service, each node could be allowed to express its own willingness to receive reward, which could be used to discriminate nodes participating in an ad hoc network. If the total willingness to receive reward of relay nodes along the route is less than the charge, the connection will be successful. This paper proposes two routing algorithms to select the route by minimizing the total willingness to receive reward. One algorithm minimizes only the total willingness to receive reward, and another algorithm minimizes both the total willingness to receive reward and the hop count. The relation between the charge and the connection success probability, and that between the charge and the benefit of relay nodes are shown.

Recently, there is a remarkable growth in IP broadcast services, and various services have been provided. There are flat-rate and time-based pricing schemes, and it is an issue to obtain the optimal price depending on the pricing scheme and the services provided. An optimal price should exist for IP broadcast services based on several viewpoints such as maximizing the revenue of IP broadcast service provider, maximizing network efficiency, maximizing user's utility, etc. This paper assumes flat-rate pricing scheme for VOD type IP broadcast services, and shows the optimal price from the viewpoint of maximizing the revenue of IP broadcast service provider. Using the user model derived from questionnaire survey, the relation between revenue and price for each server capacity is obtained by simulation. The optimal price for each server capacity is shown. The dependence of the revenue on the server capacity is also shown.

Communication services using federated personal servers have been proposed. These services have merits to manage users who access personal servers and their own contents by using personal servers. This paper proposes a thread tool which behaves like bulletin boards as a communication tool for communication services with federated personal servers. Then, the scheme to implement the function of thread tool using the function sets of group management is described.

Recently, there is a remarkable growth in IP broadcast services. There are flat-rate and time-based pricing schemes, and it is a issue to obtain the optimal price depending on the pricing scheme and services provided. An optimal price should exist for IP broadcast services based on several viewpoints such as maximizing the revenue of IP broadcast service provider, maximizing network efficiency, maximizing user's utility, etc. The purpose of this paper is to clarify the optimal price based on maximizing the revenue of IP broadcast service provider. Two kinds of pricing schemes, that is flat-rate pricing and time-based pricing, and three kinds of price elasticity are considered for VOD type IP broadcast service. The relation between revenue and price for each scheme and each price elasticity are shown. From the simulation results, the optimal price that maximizes the revenue of IP broadcast service provider is obtained.

In ad hoc network, it is necessary to increase the incentive for relaying. One idea is to give a reward for relaying, and it has been studied. The authors have discussed the relation between the willingness to receive reward and the residual battery power. In this paper, the curves obtained by the questionnaire survey are classified into several patterns. Most patterns have nonlinear shapes. For most users, the increase rate of willingness to receive reward increases according to the increase of battery power consumption (opposite of residual battery power). And, the willingness to receive reward depends on the user's sensitivity to the battery power consumption.

Packet contention occurs at a switching node ill optical packet switching networks. By using fibre delay line (FDL) buffers, packet contention can be avoided. However, the cost of a switching node with FDLs is high, and FDL introduces extra optical power loss. This paper proposes an effective wavelength assignment scheme with FDL buffers at a switching node. By suppressing the usage of FDLs, the required number of FDLs decreases, and the extra optical power loss decreases. It is shown that the proposed scheme is effective by computer simulations.

From the viewpoint of users' utility of content download service, it will be more effective to construct the multicast tree on link bandwidth than to construct the tree based on link cost etc. This paper proposes a routing method to construct a multicast tree based on link bandwidth. And, this paper proposes a routing method that a high-speed delivery and a low-speed delivery are done simultaneously by multiple servers. It is shown that the users' utility increases by constructing a multicast tree based on link bandwidth, compared with MST algorithm based on link cost etc. Moreover, it is shown that multi-server routing is effective from the viewpoint of users' utility.

Many-to-many communication such as teleconferencing or network game has to send and to receive large data on real time. Many-to-many multicasting is well suited for these communications since it makes efficient use of the network resources. In existing many-to-many multicast routing, mainly, there are two types of methods to construct multicast trees. One constructs different one-to-many multicast tree for every source. Since it constructs shortest-path trees for every source, the sources can send data with the lowest delay, however, the transmitting data increase. The other method constructs only one shared tree connected with all members. Though the delay becomes higher in this method, the transmitting data decrease. In this paper, we propose a routing algorithm using multiple static trees which can make low delay and low transmitting data many-to-many multicasting. As a result of our evaluation, by using a few static trees in the proposed algorithm, the delay performance becomes almost equal to hat of the method using shortest-path trees for every source. The proposed method can reduce the transmitting data compared with the method using one shared tree.

Recently, communication services using federated personal servers are being proposed. These services have merits to manage users who access to personal servers and to own contents by using personal servers. In this paper, communication services using federated personal servers are analyzed, and the function sets of group management which are necessary to provide the services are proposed. Also, the scheme to implement the function of group management personal servers is described.

New network attacks are found in the Internet everyday. Hence, the necessity of attack detection at an early stage is also increasing. In this paper, a new network attack detection method using IP aggregation is proposed. This method uses the difference between communication patterns in normal state and that under a network attack. We analyze flow properties of each host using IP aggregation, which helps to detect anomalous communication patterns. By using this method, we can discriminate anomalous communication hosts from other hosts.

In recent years, the amount of network traffic is rapidly increasing. And, the personal information is flowing out by virus, which makes some P2P traffic control necessary in order to maintain the efficient and stable network. The P2P traffic discrimination technique based on decryption of payload part has already been available. It has some problems, such as processing load and protection of personal information. The authors have proposed a discrimination technique of Pure P2P file sharing traffic using only header information, without decoding payload part of monitored packets. However, we still have an issue of Pure P2P discrimination between Pure P2P and other P2P. In this paper, we pay our attention to the mutual connections, and characterize the Pure P2P file sharing traffic using interval time between forward and backward connections. Moreover, we propose a new discrimination technique of Pure P2P file sharing traffic by making use of approachability of mutual connections by short time calculation. In addition, this paper verifies the effectiveness of the proposed technique.

Recently, watching image contents becomes easier using portable terminals. However, users' evaluation for watching image during movement has not been studied in detail. In this paper, we assume that the extra factors influence users' evaluation when users are watching image contents using portable terminals during movement. Then, the assumption is proved to be true by users' subjective evaluation experiment and multivariate analysis. As a result, the extra factors that influence user's evaluation are shown.

Recently, watching image contents becomes easier using portable terminals. However, users' evaluation for watching image during movement has not been studied in detail. In this paper, we assume that the extra factors influence users' evaluation when users are watching image contents using portable terminals during movement. Then, the assumption is proved to be true by users' subjective evaluation experiment and multivariate analysis. As a result, the extra factors that influence user's evaluation are shown.

Recently, differentiated services are going to be used in telecommunication networks. The price of high quality class will be high, and the price of low quality class will be low. The traffic of each class depends on the price. Therefore, in order to analyze or to design the network, it is necessary to use both teletraffic theory and economics. In this paper, we introduce the pricing method for maximizing users' utility and provider's revenue. The prices are optimized to maximize the revenue in the guaranteed bandwidth service. Disaggregate behavioural model is applied to the user's behaviour, and the relation between the user's utility and the price is obtained by simulation.

In recent years, the amount of network traffic is rapidly increasing, which makes some traffic control necessary in order to maintain the efficient and stable network. Thus, the discrimination technique of Pure P2P traffic which occupies a large portion of present network traffic is important. Pure P2P file sharing traffic is dominant in the P2P traffic, and has become a hotbed of viruses and copyright problems. The authors have proposed a discrimination technique of Pure P2P file sharing traffic using only header information, without decoding payload part of monitored packets. In this paper, we propose a new discrimination technique of Pure P2P file sharing traffic by making use of new concepts of connection direction and Mutual Connections, for the purpose of further improvement of discrimination technique the authors have proposed. In addition, this paper verifies the effectiveness of the proposed technique by comparing it with the discrimination technique based on payload decipherment.

In ad hoc networks, relay nodes use their limited resources such as battery power, CPU power, buffer, etc. for other nodes' communications. This fact discourages a user to join the ad hoc network and to become a relay node. However, if a user can receive a reward for relay, he will be encouraged to join. This research focuses on battery power consumption, residual battery power, and the time while AC power supply is not available. And, the willingness to receive reward for relay in ad hoc networks was investigated by questionnaire survey. In this paper, the relation between such factors and the willingness to receive reward for relay is analysed quantitatively. In particular, some models of willingness to receive reward for relay are estimated.

Recently, various threats appear in network. In this paper, a network attack detection method using IP aggregation is proposed. Each network attack has a peculiar communication pattern. Therefore, there is a difference between network attack communication pattern and normal communication pattern. We analyse communication patterns at every host collected by IP aggregation, and then detect network attacks. By using this method, we can discriminate worm-infected hosts from other hosts.

In core based multicast routing, the location of the core affects the efficiency of the multicast tree. This paper analyzes core based single-rooted many-to-many multicast tree by using multiple linear regression analysis, and examines how the best location of the core and its characteristics change when the number of multicast members changes. As a result, it is shown that there are four main indices which affect the delay performance of the tree. They are the maximum hop counts to any node from the core, the average of hop counts to all nodes from the core, the variance of hop counts to all nodes from the core, and the variance of the value of (node degree)/(hop counts from the core) of all nodes except the core. The result also shows that the index which mostly affects the performance of the tree depends on the number of multicast members.

Requirement for powerful ultra high-performance computer system pushes the research and development of technologies for large-scale distributed grid systems. Besides the computer systems, networks play an important role, as well. With an efficient underlying network in terms of bandwidth and delay, the performance of the whole system will be improved significantly. In the previous researches, for distributed parallel computations, for example, using Message-Passing Interface (MPI), normally the underlying network is given, then process mapping problem is resolved. In this paper, we propose an approach to design an optimized underlying lightpath network combined with a processes/processors mapping to support the large-scale parallel computing applications.

Recently, we can watch image contents such as news, movies, etc. anywhere by the ubiquitous telecommunication technology. The user's satisfaction to watch image contents has been evaluated by image quality and sound quality. However, in case of mobile terminal, the user's satisfaction will depend on the purpose and the place to watch the contents. This paper studies the users' preference to watch image contents by mobile terminals. The relation between the purpose and the place is analyzed by factor analysis. It is shown that the users' preference has 5 factors.

TCP/IP通信において、パケットの第4層以下の情報を統計的に処理することによって得られる特徴量から、第5層以上の情報により定義されるパケット分類を推定する弁別手法について、我々がこれまでに行ってきた提案とその評価について紹介する。また、分散的に構成され、かつ明確な弁別が困難とされるP2Pネットワークの例としてWinnyをとりあげ、閉鎖ネットワーク内でWinnyのみによるP2Pネットワークを構成してノイズのないデータを採取する試みを行ったのでそれについても紹介する。

Recently, we can watch image contents such as news, movies, etc. anywhere by the ubiquitous telecommunication technology. The user's satisfaction to watch image contents has been evaluated by image quality and sound quality. However, in case of mobile terminal, the user's satisfaction will depend on the purpose and the place to watch the contents. This paper studies the users' preference to watch image contents by mobile terminals. The relation between the purpose and the place is analyzed by factor analysis. It is shown that the users' preference has 5 factors.

携帯形の移動通信端末の発展により,どのような場所でも映像コンテンツの視聴が可能となりつつある.これまで,映像視聴に対するユーザの満足度は,画質や音質などの指標を用いて評価されてきたが,移動通信端末での視聴を考えた場合,ユーザの視聴目的や視聴場所によっては,映像品質のユーザ評価が異なると予想される.本稿では,携帯形の移動通信端末を用いた映像の視聴におけるユーザ志向について検討し,視聴目的と視聴場所の組合せとそれに応じたユーザ志向を因子分析により分析している.その結果,ユーザ志向には五つの因子があることを明らかにしている.

TCP/IP通信において、パケットの第4層以下の情報を統計的に処理することによって得られる特徴量から、第5層以上の情報により定義されるパケット分類を推定する弁別手法について、我々がこれまでに行ってきた提案とその評価について紹介する。また、分散的に構成され、かつ明確な弁別が困難とされるP2Pネットワークの例としてWinnyをとりあげ、閉鎖ネットワーク内でWinnyのみによるP2Pネットワークを構成してノイズのないデータを採取する試みを行ったのでそれについても紹介する。

In this paper, We introduce packet discrimination method and its evaluation which we have proposed and presented. The method enables to speculate TCP/IP packet class denned by information embeded in layer-5 or above with using information in layer-4 or below. We also present our experiment performed in a closed network, which includes pure Winny P2P network to get noiseless data generated by a representative P2P application in Japan.

In ad hoc networks, the relay nodes use their limited resources such as battery power, CPU power, buffer, etc. for other nodes' communications. This fact discourages a user to join the ad hoc network and to become a relay node. However, if a user can receive a reward for relay, he will be encouraged to join. In this paper, willingness to receive reward for relay in ad hoc networks is investigated by questionnaire survey. In particular, this paper focuses on battery power consumption, residual battery power, and the time while AC power supply is not available. Then, the relation between such factors and willingness to receive reward to relay is analysed.

In content download services, it is ideal to deliver contents to all users, whether the network is congested or not. However, in the real network services, there are a lot of situations where the servers or links are congested, and it is difficult to deliver contents to users immediately. On these situations, it is effective to reject users' requests to avoid extreme load. On the other hand, if the server rejects users' requests frequently, users' utility will decrease, and many users will leave the service. By questionnaire survey, we investigated willingness to pay of content download service which has a possibility to reject users' requests according to network congestion. This paper analyses the results of questionnaire survey, and shows the relation between call blocking probability and willingness to pay.

Recently, IP broadcast services are becoming popular with the advance of broadband telecommunication networks. Pricing scheme is dependent on IP broadcast service providers, and most providers employ flat-rate pricing. However, time-based pricing will also be used commonly in the near future. This paper analyses time-based pricing of IP broadcast services. User behaviour in VOD type IP broadcast is divided into two viewing styles. From the simulation results, the optimal price that maximises the revenue of the IP broadcast service provider is shown.

In optical packet switching networks, packet contention occurs at a switching node whenever two or more packets try to reach the same output port of the node at the same wavelength, consequently network performance decreases. Previous works on optical packet switching flourishingly adopted wavelength conversion, optical buffering, or deflection routing to avoid packet loss. However, systems cost and control complexity is high. This paper proposes an effective wavelength assignment scheme to decrease packet contention. It is shown that the proposed scheme is effective through computer simulations.

The test-driven software development method will make specification change easy compared with the conventional water fall model. The software release of the former will be done shortly compared with the latter. However, the quality assurance is difficult, because it is necessary to extract the test case without the source code. Thus, an algorithm is proposed to extract the test case from the specification, and the test case extraction tool using this algorithm is also proposed. It is shown that this tool can extract test cases for the state-less model appropriately. This paper considers whether the proposed method can be applied to the state transition model to extract test cases of actual SIP media server. The result shows that the test cases can be extracted appropriately by considering two or more possibilities for one item to estimate the output for each input.

In recent years, development of the technique of discriminating PureP2P traffic by network traffic observation is progressing. We paid our attention to the technique not from payload part but from header information. We have shown possibility that discrimination by "New Connection Rate" is useful. In order to apply "New Connection Rate" to general PureP2P, the elucidation of the principle based on the characteristic of PureP2P traffic is important. Then, we performed the file-sharing simulation using 200 machines in the Hokuriku IT Development Support Center in order to solve the principle of traffic by PureP2P file-sharing software. Especially, we focused on the influence of the network constitution of P2P to traffic characteristics, and paid our attention to the existence of initial nodes and band composition ratio in a network. This showed that the existence of initial nodes and the band composition ratio caused the increase in the number of connection nodes.

This study provides a solution for discriminating P2P traffic below transport layer and handling the traffic appropriately.

If a static multicast tree is used for one-to-many multicast, the most efficient tree depends on the characteristics of the receivers such as the number of receivers. And, the efficiency of the tree has already been analyzed. However, as for many-to-many multicast, the efficiency of the tree has not been analyzed. This paper analyzes how various components affect the efficiency of the many-to-many multicast tree. As a result, it is shown that the most efficient many-to-many multicast tree does not depend on the number of multicast members. The performance of the tree depends mostly on the diameter of the tree, the variance of node degrees, and the number of leaf nodes.

Given a static traffic condition, the motivation of logical topology design in IP over WDM networks is to optimize the network resource utilization to improve network performance. When it is necessary, the network can dynamically change its logical topology corresponding to the changing traffic conditions. Besides the traffic pattern changing, other network events may also need the logical topology to be reconfigured, such as the physical topology changes. Here, a faster near optimal reconfiguration can keep up with the changes of the network and bring a better system utilization compared to the long running time optimization for an exact solution. A fast reconfiguration needs a fast logical topology design tool. In this paper, a Tabu search (TS) method to resolve the logical topology design problem is proposed. The motivation is to find a feasible tool of logical topology design for network operators to re-optimize the network utilization within a shorter endurable computation time.

近年, インターネットのブロードバンド化に伴い大容量のコンテンツ配信サービスが注目されてきている.また, ユーザのニーズが多様化していることから, 優先制御を用いて複数クラスの品質を提供する差別化サービスが新たなサービスとして検討されている.このようなサービスでは, ユーザのニーズを予測し, それに合わせたサービスを提供することが非常に重要である.本稿では, 優先制御コンテンツ配信におけるユーザの行動を支払意思額によりモデル化し, ユーザ便益・事業者収益を最適とする条件をシミュレーションにより評価している.その結果, ユーザのための最適な料金設定と, 事業者のための最適な料金設定が異なることを明らかにしている.

近年, インターネットのブロードバンド化に伴い大容量のコンテンツ配信サービスが注目されてきている.また, ユーザのニーズが多様化していることから, 優先制御を用いて複数クラスの品質を提供する差別化サービスが新たなサービスとして検討されている.このようなサービスでは, ユーザのニーズを予測し, それに合わせたサービスを提供することが非常に重要である.本稿では, 優先制御コンテンツ配信におけるユーザの行動を支払意思額によりモデル化し, ユーザ便益・事業者収益を最適とする条件をシミュレーションにより評価している.その結果, ユーザのための最適な料金設定と, 事業者のための最適な料金設定が異なることを明らかにしている.

Recently, large-size contents are delivered to users by content delivery service through broadband networks. The necessity for the service which guarantees quality is increasing in the Internet, because users' requirements are diversified. In such services, it is important to predict users' needs and to provide the services matching up to it. In this paper, the user's behaviour in content delivery service with priority control is modeled by using willingness to pay, and the users' benefit and the provider's revenue are maximized. As a result, it is shown that the optimal pricing for users and that for the provider are different.

Multicast technology is expected to become widely used, since its link sharing capability makes it well suited to large size content delivery services. In multicast content delivery services, the speed of content delivery is adjusted to the slowest user's speed in the multicast tree. Therefore, it is effective to separate slow users from fast users according to users' available rate. This paper compares the users' utility of the delivery with multiple speeds and that of the delivery with blocking slow link users. It is shown that the delivery with multiple speeds is more effective when the users are not so sensitive to the waiting time to get the content.

HMIPv6 (Hierarchical Mobile IPv6) protocol introduces MAP (Mobility Anchor Point) which relays all packets between HA (Home Agent) and MN (Mobile Node). This method can reduce the amount of signaling traffic and improve the handover speed. Since very high loads are concentrated in a MAP and the load of MAP possibly influences the communication quality, the load balancing method by suitable policy is needed in multi-layered MAP network. This paper proposes the load balancing method with score calculated according to attributes of mobile nodes, such as handover frequency and holding time. Simulation results show that this proposed method can reduce the load of whole network while achieving the load balancing on multi-layered MAP in HMIPv6 network.

User's demand of the quality of streaming service depends on the intrinsic characteristics of the content. Various things can be considered as the characteristic. In this paper, it is considered that intensity based on sound or image quality is characteristics. The subjective evaluation experiment was done for several kinds of content. The relation between content characteristics and content types is shown by using principal component analysis. Then, based on the image quality and the sound quality, the contents are divided into several groups by cluster analysis.

In network management, decisions are based on online analysis of monitoring data. In order for such decisions to be effective, analysis of monitoring data should have high validity. Unfortunately, due to the time constraints complicated analysis of large volumes of performance data is impossible. In this paper we introduce a variable that estimates the certainty of analysis of limited data volumes. The proposed certainty variable is tested with various sources of monitoring data.

With the rapid development of the Internet, the P2P communication technology which does not use any servers has kept advancing apace. However, the progress of the P2P network has lead to the illegal file sharing in addition to the rapid growth of network traffic. Thus, this research aims to discriminate users of pureP2P file sharing application from other users. This paper proposes some features including "new connection rate" which are useful for the P2P user discrimination. To automate discrimination processing, SVM(Support-Vector-Machine) is applied to this discrimination, and its accuracy is evaluated based on some experiments.

Although management paradigms have lived through a number of major changes in the last 10 years, most of them are not proactive and very weakly depend on current state of managed network domain. Up to the day, monit oring and management activities are executed separately and quite often have independent infrastructures within the same network domain. This paper looks into the possibility of using online monitoring data, both passive. as well as obtained by active measurements, in order to perform management activities online. The usability of various sources of data are tested by neural network predictions.

Mobile IPv6 allows a mobile node to move from one link to another without changing the mobile node's address. And the Hierarchical Mobile IP Mobility Management (HMIPv6) proposes to reduce the amount of signallinig traffic and improve the performance of Mobile IPv6 in terms of handoff speed. HMIPv6 allows multiple Mobility Anchor Points located at any level in hierarchical network. However, there are no effective method to determine the MAP which a mobile node use. This paper presents optimal MAP selection method based on the attribute of mobile nodes. Simulation results show that this method greatly reduces the amount of signallinig traffic and achieve the load balancing in Hierarchical Mobile IPv6 network.

WDM optical networks represent the direction to the future high capacity access network applications. Here, we investigate a single-hop WDM star network. It is based on a passive star-coupled configuration. In this paper, we investigate the influence of long tuning latency on the performance of one slow tunable transmitter and multiple fixed receivers architecture. Assuming the existence of traffic pattern in certain time period, two heuristic dynamic channel allocation algorithms based on distributed traffic analysis are proposed. To investigate the performance, evaluation through simulations is done in terms of network throughput and packet delay. It is shown that the proposals are efficient to optimize the wavelength channel utilization.

In multicast contents delivery services, the waiting time of users will decrease if the contents delivery interval becomes shorter. However, if we set all contents' delivery intervals too short, it will cause network congestion. Therefore, it seems effective to set short delivery intervals to popular contents and set long delivery intervals to unpopular contents. In this paper, we show it is able to increase users' utility to set delivery interval based on each content's popularity. And, we noticed that setting delivery interval based on each content's popularity is more effective in situations where many users want to obtain contents as quickly as possible.

WDM optical networks represent the direction to the future high capacity access network applications. Here, we investigate a single-hop WDM star network. It is based on a passive star-coupled configuration. In this paper, we investigate the influence of long tuning latency on the performance of one slow tunable transmitter and multiple fixed receivers architecture. Assuming the existence of traffic pattern in certain time period, two heuristic dynamic channel allocation algorithms based on distributed traffic analysis are proposed. To investigate the performance, evaluation through simulations is done in terms of network throughput and packet delay. It is shown that the proposals are efficient to optimize the wavelength channel utilization.

Recently, much attention has been given to high-speed optical networks as the Internet becomes popular and has large traffic. It is now expected to realize the optical network techniques : wavelength routing, optical burst switching, and optical packet switching. Considering the characteristics of these technologies, it is more desirable that users can choose various switching systems depending on the provided contents. For example, wavelength routing that is circuit switching system in optical network is most suitable for real-time data. This paper proposes not only effective but also high-quality content delivery method that switches both wavelength routing and optical burst switching according to the types of contents. The performance of the proposed method is evaluated by computer simulation.

In multicast contents delivery services, the waiting time of users will decrease if the contents delivery interval becomes shorter. However, if we set all contents' delivery intervals too short, it will cause network congestion. Therefore, it seems effective to set short delivery intervals to popular contents and set long delivery intervals to unpopular contents. In this paper, we show it is able to increase users' utility to set delivery interval based on each content's popularity. And, we noticed that setting delivery interval based on each content's popularity is more effective in situations where many users want to obtain contents as quickly as possible.

WDM optical networks represent the direction to the future high capacity access network applications. Here, we investigate a single-hop WDM star network. It is based on a passive star-coupled configuration. In this paper, we investigate the influence of long tuning latency on the performance of one slow tunable transmitter and multiple fixed receivers architecture. Assuming the existence of traffic pattern in certain time period, two heuristic dynamic channel allocation algorithms based on distributed traffic analysis are proposed. To investigate the performance, evaluation through simulations is done in terms of network throughput and packet delay. It is shown that the proposals are efficient to optimize the wavelength channel utilization.

Recently, much attention has been given to high-speed optical networks as the Internet becomes popular and has large traffic. It is now expected to realize the optical network techniques : wavelength routing, optical burst switching, and optical packet switching. Considering the characteristics of these technologies, it is more desirable that users can choose various switching systems depending on the provided contents. For example, wavelength routing that is circuit switching system in optical network is most suitable for real-time data. This paper proposes not only effective but also high-quality content delivery method that switches both wavelength routing and optical burst switching according to the types of contents. The performance of the proposed method is evaluated by computer simulation.

Differentiated services, which have multiple classes using priority control, have recently been introduced as a new service. In this paper, the prices are optimized to maximize the revenue in the guaranteed bandwidth service with flat monthly rate. Disaggregate behavioural model is applied to the user's behaviour, and the relation between the user's utility and the price is obtained by simulation. The optimum pricing function is estimated as a power function.

Soft release in short term can be expected to be achieved by applying the test-driven development technique to communication software. However, it is necessary to extract test items in the situation before coding programs to be tested. The test items are extracted by three steps ; covering extraction of input system, forecast of output, and generation of test program. Especially, the first step is difficult. To solve this problem, this paper proposes an algorithm that pulls out parameters from specifications and extracts the input system by paying attention within range where parameters can be taken. Applying this algorithm to the development of reverse proxy software, which has plural classes, 15 test items were extracted from class I and 37 test items were extracted from class II.

Differentiated services, which have multiple classes using priority control, have recently been introduced as a new service. In this paper, the prices are optimized to maximize the revenue in the guaranteed bandwidth service with flat monthly rate. Disaggregate behavioural model is applied to the user's behaviour, and the relation between the user's utility and the price is obtained by simulation. The optimum pricing function is estimated as a power function.

Soft release in short term can be expected to be achieved by applying the test-driven development technique to communication software. However, it is necessary to extract test items in the situation before coding programs to be tested. The test items are extracted by three steps ; covering extraction of input system, forecast of output, and generation of test program. Especially, the first step is difficult. To solve this problem, this paper proposes an algorithm that pulls out parameters from specifications and extracts the input system by paying attention within range where parameters can be taken. Applying this algorithm to the development of reverse proxy software, which has plural classes, 15 test items were extracted from class I and 37 test items were extracted from class II.

WDM optical networks represent the direction to future high capacity wide-area network applications. By creating optical paths between nodes in the core networks, logical topology can be created over the physical topology. In this paper we consider the method of designing logical topology in wavelength routed IP networks. For a network operator to achieve logical topology design especially for the fast restoration from large network failures after disasters such as earthquake where logical topology reconfiguration is required, comparing to the performance requirement the computing time of a feasible solution is critical. The exact formulation of the logical topology problem is usually given as Mixed Integer Linear Programming, but it grows intractable with increasing network size. For a fast design we have proposed a new simulated annealing approach to determine a better logical topology within a shorter time. For further study on fast logical topology design using meta-heuristics, here we compare simulated annealing with a genetic algorithm. Through performance comparisons, we show that our proposed simulated annealing approach can design a better logical topology in a shorter time comparing to previous simulated annealing and genetic approaches.

WDM optical networks represent the direction to future high capacity wide-area network applications. By creating optical paths between nodes in the core networks, logical topology can be created over the physical topology. In this paper we consider the method of designing logical topology in wavelength routed IP networks. For a network operator to achieve logical topology design especially for the fast restoration from large network failures after disasters such as earthquake where logical topology reconfiguration is required, comparing to the performance requirement the computing time of a feasible solution is critical. The exact formulation of the logical topology problem is usually given as Mixed Integer Linear Programming, but it grows intractable with increasing network size. For a fast design we have proposed a new simulated annealing approach to determine a better logical topology within a shorter time. For further study on fast logical topology design using meta-heuristics, here we compare simulated annealing with a genetic algorithm. Through performance comparisons, we show that our proposed simulated annealing approach can design a better logical topology in a shorter time comparing to previous simulated annealing and genetic approaches.

Multicast communication is a remarkable technology that can substantially decrease network traffic and can save network resources and transmission costs. In multicasting, the transmission speed must be set to the lowest one among the available capacities of links in the multicast tree. This characteristic increases the data transmission time, and accordingly the content could not be delivered efficiently. We have studied an adaptive delivery system to dynamically select multicasting and store-and-forward transferring. This system was proved as an effective system in best-effort networks. In this paper, we put remarkable points on transmission control and a store-and-forward fanction, after considering link capacities. Experimental results show that the implemented system realizes effective data delivery with expected behavior in a test network, we have implemented application-assisted adaptive multicast with transmission control.

マルチキャストはネットワークのトラヒックを減らし, ネットワークリソースと通信コストを節約することのできる非常に優れた技術である.マルチキャストを使用する場合, その通信速度はマルチキャストツリー内で最小の使用可能帯域に合わせる必要がある.この特徴はデータの配信時間を増大させ, コンテンツの効果的な配信を妨げるものである.筆者らは途中ルータに搭載したアプリケーションによって, マルチキャストと蓄積配信を動的に選択する適応形マルチキャストシステムの開発を行っている.このシステムはベストエフォートのネットワークにおいて効果的であることが証明されている.本稿では, ネットワークの輻輳状況を判断した後の送信制御機能および蓄積再配信機能に重点をおいた開発を行い, 作成したアプリケーションの動作確認を行っている.その結果, 実験ネットワークにおいて, 適切に送信制御された蓄積配信を実現している.

Multicast communication is a remarkable technology that can substantially decrease network traffic and can save network resources and transmission costs. In multicasting, the transmission speed must be set to the lowest one among the available capacities of links in the multicast tree. This characteristic increases the data transmission time, and accordingly the content could not be delivered efficiently. We have studied an adaptive delivery system to dynamically select multicasting and store-and-forward transferring. This system was proved as an effective system in best-effort networks. In this paper, we put remarkable points on transmission control and a store-and-forward fanction, after considering link capacities. Experimental results show that the implemented system realizes effective data delivery with expected behavior in a test network, we have implemented application-assisted adaptive multicast with transmission control.

Multicast communication is a remarkable technology that can substantially decrease network traffic and can save network resources and transmission costs. In multicasting, the transmission speed must be set to the lowest one among the available capacities of links in the multicast tree. This characteristic increases the data transmission time, and accordingly the content could not be delivered efficiently. We have studied an adaptive delivery system to dynamically select multicasting and store-and-forward transferring. This system was proved as an effective system in best-effort networks. In this paper, we put remarkable points on transmission control and a store-and-forward fanction, after considering link capacities. Experimental results show that the implemented system realizes effective data delivery with expected behavior in a test network, we have implemented application-assisted adaptive multicast with transmission control.

都市部におけるMANは格子形であることが多い.そのようなMANにおけるスパニングツリーの形状としては, 障害復旧の観点を重視すると, 渦巻形が適している.本稿では, MANの格子形構造の特徴に着目した表現方法を定義し, 多面構成された渦巻形スパニングツリーの形状の類似性を比較する.それによりリンク障害の影響度や冗長面確保のための類似度を評価する管理手法を提案し, 渦巻形スパニングツリーによる多面構成格子形MANの評価を行う.

都市部におけるMANは格子形であることが多い.そのようなMANにおけるスパニングツリーの形状としては, 障害復旧の観点を重視すると, 渦巻形が適している.本稿では, MANの格子形構造の特徴に着目した表現方法を定義し, 多面構成された渦巻形スパニングツリーの形状の類似性を比較する.それによりリンク障害の影響度や冗長面確保のための類似度を評価する管理手法を提案し, 渦巻形スパニングツリーによる多面構成格子形MANの評価を行う.

WDM optical networks represent the direction to the future high capacity access network applications. Here, we investigate a single-hop WDM star network. It is based on a passive star-coupled configuration. Employing slow tunable transceivers means low system cost, in this paper we investigate the influence of long tuning latency on the performance of one slow tunable transmitter and multiple fixed receivers architecture. Assuming the existence of traffic pattern in certain time period, two heuristic optimal channel allocation algorithms for slotted ALOHA protocol are proposed. To investigate the performance, evaluation through simulations is done in terms of network throughput, packet delay. We show that the proposals are efficient to optimize the wavelength channel utilization.

WDMシングルホップネットワークの構成要素として,各ノード一つの低速波長切換送信機と,各ノード波長数分の固定波長受信機を仮定する.これらのデバイスはいずれも安価なものである.このようなネットワークにおいて,トラヒック変動が起こったときは,波長割当を変更することになる.波長の変更に要する時間は送信機を使用できないので,波長切換を行うノード数がなるべく少なくなるように波長割当を行うと,スループット及び遅延特性が改善される.本稿では,その方式を提案し,特性を求めている.

To offer the guaranteed multiple-priority services, we should know what quality is needed and what price is appropriate. The willingness to pay (WTP) would change according to the maximum bandwidth which can be used during non-busy period, even if the minimum guaranteed bandwidth is fixed. This paper shows the relation between the maximum bandwidth and the willingness to pay of multiple-priority classes, by subjective evaluation experiment and questionnaire survey. Using the WTP, the price of the minimum guaranteed bandwidth is quantified.

近年,優先制御を用いて複数クラスの品質を提供する差別化サービスが新たなサービスとして検討されており,品質に対する料金設定が問題となっている.本稿では,二つのアプリケーションの帯域幅による品質の違いを提示し,それぞれの最低保証帯域幅に対する支払意思額や,サービス品質に対する満足度を主観評価実験により調査した.そして,最低保証帯域幅の主観品質評価と支払意思額の関係を定量的に明らかにした.

In the download type contents delivery service using multicast, the delivery speed must be adjusted to the minimum bandwidth of the multicast tree. The user's utility depends on the waiting time, and the waiting time depends on the minimum bandwidth. Therefore, from the viewpoint of user's utility, it will be effective to block slow link user's call by using call admission control (CAC). However, by using CAC, the total number of users will decrease, and the total users' utility may decrease. This paper studies the relation between the delivery interval and the total users' utility, and that between the blocking probability and the total users' utility.

In order to build MAN, it is important to consider existing building structures. By using spanning tree for broadcast communication on all-optical MAN, it will be possible to improve reliability. Spiral spanning tree is fit for this purpose. The restoration procedure is very simple, and it is fast to recover from link failure. This method can be applied to multi-layered configuration using WDM technology for future enhancement.

In present mobile communications, hand-over may often cause packet loss andre-ordering. In case of multicast this can be a serious problem, as lost UDP packets are not retransmitted. This is also the reason of considerable deterioration of quality in video streaming services. This paper is considering a method of avoiding packet loss. It is based on existing HMIPv6(Hierarchical Mobile IPv6)+FMIPv6(Fast Handover for Mobile IPv6) methods. The complete extermination of packet loss is implemented through providing abilities to transmit packets to intermidiate router (Mobility Anchor Point: MAP), buffer them, and control multicast groups.

Recent advent of QoS requirements at higher application layers and increasing complexity of network routing technologies demand higher reliability and comprehensiveness from network performance management system. We propose an effective framework of network performance monitoring that can be vised to obtain relatively short-term properties of network performance or as a part of automatic management systems.

Contents delivery will become more popular in the near future, and its traffic is expected to become dominant in the network. Generally, the user's utility (satisfaction) decreases as the waiting time for acquisition of contents increases. The total users' utility can be maximized by using priority control in consideration of user's behaviour. This paper studies the priority control based on various types of utility function. The formula for estimating the mean waiting time is derived for priority control available in packet switching systems. The optimal condition is defined in terms of the traffic load and the waiting time for high and low priority classes, respectively. Finally, the optimal condition for priority and non-priority classes is analysed using numerical examples.

With the rapid growth of the computer network, it is becoming difficult for routers to handle growing amount of routing data. Hierarchical routing is one of the solutions to this problem. In the current hierarchical routing architecture, an administrator must decide routing groups and properly assign a static metric to each link manually. This process is complicated. Furthermore, to maintain the routing performance, dynamic metrics such as traffic should be considered in constructing hierarchical groups. This paper proposes optimization method that introduces traffic information into the decision of hierarchical routing groups. This method can be applied to a random network, and thus it can reduce the administrative process. Futhermore, this method can decide the effective hierarchical routing groups considering traffic impact. The obtained results show that our method can greatly reduce the size of routing data while keeping low average delay in the network.

Mobile IPv6 allows a mobile node to move from one link to another without changing the mobile node's address. And the Hierarchical Mobile IP Mobility Management (HMIP) proposes to reduce the amount of signallinig traffic and improve the performance of Mobile IPv6 in terms of handoff speed. HMIPv6 allows multiple Mobility Anchor Points located at any level in hierarchical network. However, there are no method to determine the number of MAPs or their location in the network. This paper presents optimal location based on the results of analysis of load under several different network topologies.

Recently, much attention has been given to high-speed optical networks as the Internet becomes popular and has large traffic. It is now expected to realize wavelength routing, optical burst switching, and optical packet switching as optical network techniques. Considering the characteristics of these technologies, it is more desirable for users to choose various transport networks according to provided services. This paper proposes an effective content delivery method that switches the three optical techniques depending on the content size. The performance of the proposed method is evaluated by computer simulation.

The increase of trasfer capacity in modern networks does not necessarily mean better performance as interactions of a large number of simultaneous flows can lead to unpredictable traffic conditions that can finally result in drastic performance deterioration. Traffic modeling itself still faces great challenges in creating the model of traffic in today's Internet, and with greately increased traffic in high-speed networks this problem becomes only worse. Thus, performance monitoring becomes the most important part of managing such networks. This paper proposes a lightweight monitoring solution by distributed active measurement that overcomes shortcomings of currently popular managment paradigms.

Measurement-based CAC Call Admission Control) is widely recognized as a scheme to guarantee QoS of VoIP. PSTN Fallback, which is a measurement-based CAC scheme, has already been implemented in routers for practical use. Scalability is a problem of PSTN Fallback, because SAA (Security Assurance Agent) probes are transmitted o all destination IP addresses. In particular, it is almost impossible to transmit probes to thousands of destination IP addresses in a large-scale network. In this paper, a hierarchical measurement-based CAC is proposed to solve the problem of scalability. Then, the probe measurement mechanism is modeled, and the measurement success probability is calculated. Furthermore, from the viewpoint of SAA probe traffic, the most suitable hierarchy is examined.

With the rapid development of communication technology, we now have various kinds of access circuits for IP connection based on Ethernet (wired or wireless), cellular phone or PHS. Since each circuit is available at relatively low cost, it is easy to use multiple circuits simultaneously. But bit error rate of some circuits can be unacceptably high under bad conditions. In this paper, we propose a new link aggregation method called LADUP (Link Aggregation by Duplication and Unification of Packets) to achieve high TCP throughput with effective use of multiple circuits. The packets duplicated in LADUP depends on the quality related conditions of the circuits and they are defined as either only ACK packets or all the packets in this paper. The application conditions of these two alternatives are demonstrated based on their throughput evaluation results in burst and random packet loss models.

SmallWorld that represents human behaviour and psychology is widely noticed. As telecommunication networks are used by human beings, SmallWorld topology will be effective for them. This paper studies the network design method using SmallWorld. First, a new topology design method is proposed. Then, the performance is compared with other topologies. It is shown that the design method using SmallWorld is effective.

To cope with the congestion generated by intensive demand of large size contents delivery, it will be effective to introduce multiple quality classes. When a user requests a content, a table of several combinations of the delivery time and the price is shown to the user, and the service class is chosen by the user appropriately. It becomes possible to decrease the load at peak usage hours efficiently, and to use the resource of off-peak hours effectively. In this paper, although the case of a single link model is examined, how to extend the proposed system to multiple link model and its effect are also shown. Furthermore, even if there is an error between the traffic pattern predicted for setting price and the actual traffic pattern, the proposed method is effective.

With the rapid development of communication technology, we now have various kinds of access circuits for IP connection based on Ethernet (wired or wireless), cellular phone or PHS. Since each circuit is available at relatively low cost, it is easy to use multiple circuits simultaneously. But bit error rate of some circuits can be unacceptably high under bad conditions. In this paper, we propose a new link aggregation method called LADUP (Link Aggregation by Duplication and Unification of Packets) to achieve high TCP throughput with effective use of multiple circuits. The packets duplicated in LADUP depends on the quality related conditions of the circuits and they are defined as either only ACK packets or all the packets in this paper. The application conditions of these two alternatives are demonstrated based on their throughput evaluation results in burst and random packet loss models.

SmallWorld that represents human behaviour and psychology is widely noticed. As telecommunication networks are used by human beings, SmallWorld topology will be effective for them. This paper studies the network design method using SmallWorld. First, a new topology design method is proposed. Then, the performance is compared with other topologies. It is shown that the design method using SmallWorld is effective.

To cope with the congestion generated by intensive demand of large size contents delivery, it will be effective to introduce multiple quality classes. When a user requests a content, a table of several combinations of the delivery time and the price is shown to the user, and the service class is chosen by the user appropriately. It becomes possible to decrease the load at peak usage hours efficiently, and to use the resource of off-peak hours effectively. In this paper, although the case of a single link model is examined, how to extend the proposed system to multiple link model and its effect are also shown. Furthermore, even if there is an error between the traffic pattern predicted for setting price and the actual traffic pattern, the proposed method is effective.

近年,インターネツトでは品質を確保したサービスの必要性が高まってきており,優先制御を用いて複数クラスのサービスを提供する差別化サービスが検討されている.差別化サービスを提供する場合,品質をどの程度確保したらよいのか,またその料金をどのように設定すればよいのかが問題となる.そこで本稿では,支払い意思額に基づくユーザ行動を用いたISP収入最大化のための最適料金設定法について検討する.

In recent years, the necessity of quality guaranteed services is increasing in the Internet. The differentiated services, which have multiple classes using priority control, are being introduced. To offer the differentiated services, we should know what quality is needed and what price is appropriate. In this paper, the user based optimum pricing method is shown for maximizing the revenue of ISP by using the willingness to pay function.

To offer the differentiated services, which have multiple classes using priority control, we should know what quality is needed and what price is appropriate. And, the willingness to pay would change according to the maximum bandwidth which can be used during non-busy period, even if the minimum guaranteed bandwidth is fixed. This paper shows the effect of the maximum bandwidth on the relation between the minimum guaranteed bandwidth and the willingness to pay of differentiated services, by subjective evaluation experiment and questionnaire survey.

To offer the differentiated services, which have multiple classes using priority control, we should know what quality is needed and what price is appropriate. And, the willingness to pay would change according to the maximum bandwidth which can be used during non-busy period, even if the minimum guaranteed bandwidth is fixed. This paper shows the effect of the maximum bandwidth on the relation between the minimum guaranteed bandwidth and the willingness to pay of differentiated services, by subjective evaluation experiment and questionnaire survey.

This study introduces Reinforcement Learning (RL) to solve the problem of Call Admission Control (CAC) and Routing for multicast connections in networks serving multiple service classes. The network system is trained to find out the optimal control policy that brings up the highest amount of reward in long-run. For a manageable solution and realizable training time, decomposition is carried out. To demonstrate the prominence of RL-based routing against MOSPF (Multicast extension to Open Shortest Path First) protocol, which is a routing protocol with high performance among available ones, we consider different criteria, including reward rate and call drop rate in simulations.

This study introduces Reinforcement Learning (RL) to solve the problem of Call Admission Control (CAC) and Routing for multicast connections in networks serving multiple service classes. The network system is trained to find out the optimal control policy that brings up the highest amount of reward in long-run. For a manageable solution and realizable training time, decomposition is carried out. To demonstrate the prominence of RL-based routing against MOSPF (Multicast extension to Open Shortest Path First) protocol, which is a routing protocol with high performance among available ones, we consider different criteria, including reward rate and call drop rate in simulations.

ネットワーク管理は,従来,ネットワークの状態監視や障害処理に限定されたものであった.しかし,現在は,要求が多様化して,ネットワーク内のさまざまなパラメータの監視が必要となり,またそれはルーチング機能などとも関わってくるようになってきている.SNMPは極めて単純にできており,メーカが工場で製造する段階でルータに組み込むことができる標準化された限られたMIBのリストを利用している.しかしながら,SNMPでは,MIBの遠隔監視(RMON)において,大量の監視トラヒックが生じ,また集めた情報を集中処理することがボトルネックになることもある.このような問題は,MIBに基づく監視を行うことにより生ずるものである.そこで,エンドツーエンドでの動的な測定により必要な情報を得る方法があり得る.この方法を用いれば,MIBを用いる方法と比べて,トラヒックが減少して問題点が解決し,より柔軟にネットワークのオペレーションや管理を行える可能性がある.本稿は,その可能性の検討を行ったものである.

ネットワーク管理は,従来,ネットワークの状態監視や障害処理に限定されたものであった.しかし,現在は,要求が多様化して,ネットワーク内のさまざまなパラメータの監視が必要となり,またそれはルーチング機能などとも関わってくるようになってきている.SNMPは極めて単純にできており,メーカが工場で製造する段階でルータに組み込むことができる標準化された限られたMIBのリストを利用している.しかしながら,SNMPでは,MIBの遠隔監視(RMON)において,大量の監視トラヒックが生じ,また集めた情報を集中処理することがボトルネックになることもある.このような問題は,MIBに基づく監視を行うことにより生ずるものである.そこで,エンドツーエンドでの動的な測定により必要な情報を得る方法があり得る.この方法を用いれば,MIBを用いる方法と比べて,トラヒックが減少して問題点が解決し,より柔軟にネットワークのオペレーションや管理を行える可能性がある.本稿は,その可能性の検討を行ったものである.

Network management has been only for network troubleshooting and status-reading. Requirements evolved and now management is in very close relation with monitoring of various parameters in the network, including its core routing functionality. SNMP was devised to be extremely simple, with a limited standardized list of MIBs that vendors can implement in routers at the factory. However, due to the nature of SNMP that did not change its operation since the emergence of remote monitoring MIBs (RMON), monitoring produces considerable traffic, and bottleneck both in the network and at processing center. There is a possibility that low performance of MIB-based monitoring can be improved by stand-alone end-to-end active measurement, that will cause little overhead, offer much more flexible control over operation, and can serve as a basis for creating intelligent event-driven management. This paper looks studies this possibility.

近年,インターネットでは品質を確保したサービスの必要性が高まってきており,優先制御を用いて複数クラスのサービスを提供する差別化サービスが検討されている.差別化サービスを提供する場合,品質をどの程度確保したらよいのか,またその料金をどのように設定すればよいのかが問題となる.本稿では,ストリーミング配信用コンテンツを使用した主観評価実験と,ダウンロードの待ち時間を提示したアンケート調査により,差別化サービスにおける最低保証帯域とその帯域に対する支払い意思額の関係を求めている.

In recent years, the necessity of quality guaranteed services is increasing in the Internet. The differentiated services, which have multiple classes using priority control, are being introduced. To offer the differentiated services, we should know what quality is needed and what price is appropriate. In this paper, the quantitative relation between the guaranteed minimum bandwidth and the willingness to pay is shown by the subjective experiment of streaming contents and by the questionnaire survey of the waiting time for download.

In recent years, the necessity of quality guaranteed services is increasing in the Internet. The differentiated services, which have multiple classes using priority control, are being introduced. To offer the differentiated services, we should know what quality is needed and what price is appropriate. In this paper, the quantitative relation between the guaranteed minimum bandwidth and the willingness to pay is shown by the subjective experiment of streaming contents and by the questionnaire survey of the waiting time for download.

In recent years, various kinds of circuits such as wired LAN, wireless LAN, cellular phone and PHS have become available due to the rapid development of communication technology. Since there is such a large choice of circuits, multiple communication circuits can be used simultaneously easily. The objective of this paper is to propose and study a method to transfer duplicated data packets through multiple physical circuits of various quality in order to construct high throughput and robust network system.

In Mobile IPv6, the binding cache is updated by the binding update sent by the mobile node. The binding update is sent to the home agent and the present correspondent node. Most correspondent nodes do not know the care of address of the mobile node before starting communication. Therefore, most correspondent nodes have to send the packets to the home agent of the mobile node at first. This paper proposes to extend the binding cache function. In the proposed scheme, the mobile node sends the binding update to the home agent, and the home agent re-sends it to some agents which the mobile node often communicates with. By this, most correspondent nodes do not have to send the packets to the home agent at first, and they can send the packets directly to the care of address of the mobile node. Then, the delay of the first packet decreases, and it is fit for delay critical services such as VoIP.

Multicast is going to be popular as a multipoint communication mechanism for effective transmission of network traffic. However, conventional telecommunication networks are not suited for multicast since it is supposed that only unicast traffic is transmitted. It is therefore important to consider multicast network design. On the other hand, SmallWorld that represents human behaviour and psychology is widely noticed. As telecommunication networks are used by people, SmallWorld topology will be effective for them. This paper examines the costs for network construction in case where SmallWorld is applied to the future network that supports both multicast and unicast traffic.

コンテンツ配信において,その効用は待ち時間に対し非線形に減少する.また,コンテンツ取得までの待ち時間に対するユーザ効用は,コンテンツの属性に応じて異なると予想される.本稿では,待ち時間感度の違うコンテンツを配信した場合の多クラス優先制御によるユーザ効用の最大化について検討する.

In the contents delivery service, the user's utility depends on the waiting time to get the contents. According to the former studies, the utility decreases nonlinearly in accordance with the increase of the waiting time. In the contents delivery network, various kinds of contents are delivered, and they have various attributes. The decreasing rate of the utility against the increase of the waiting time will depend on the content attribute. In this paper, a multi-class priority control based on the waiting time sensitivity is proposed in order to make the total users' utility maximum.

本稿は,光波長ルーチングネットワークにおける論理トポロジーの再構築手法について検討を行っている.動的に変化するトラヒックに対して,現時点でのトラヒックに大きな混乱を与えないような論理トポロジーの再構築が必要とされている.頻繁に再構築が行われる際,より短時間で論理トポロジーを設計する必要がある.多くの場合混合整数線形計画法から得ることができるが,厳密に論理トポロジーを形成すれば,ネットワークが大きくなるにつれて急速に計算量が増大し,計算処理は困難になる.そこで,メタヒューリスティック手法を用いて,より少ない計算回数で論理トポロジー構築を行う高速焼きなまし手法を提案している.

Multicast communication is a remarkable technology that can substantially decrease network traffic and can save network resources and transmission costs. In multicasting, the transmission speed must be set to the lowest one among the available capacities of links in the multicast tree. This characteristic increases the data transmission time, and accordingly the content could not be delivered efficiently. We have studied an adaptive delivery system to dynamically select multicasting and store- and-forward transferring. This system was proved as an effective system in best-effort networks. This paper proposes the implementation methods of the adaptive content delivery system fit for the Internet considering the current IP multicast routing protocol.

コンテンツ配信ネットワークでは,大容量コンテンツの配信時に,集中的な輻輳が発生することが予想される.本稿では,この問題に対処する方法として,配信待ち時間に関して複数品質クラスを設け,料金に差を付ける方式について検討している.料金は優先度に応じて設定し,更にピーク時には各クラスの料金をより高くする.特に,優先度の高いクラスの料金は低いクラスよりも大きく上げる.ユーザに,必要の度合いに応じて適切に品質クラスを選択させることで,ピーク時の負荷をオフピーク時に誘導し,一日のトラヒックを平準化することが可能となる.本稿では,ユーザの効用の観点からこのシステムの効果を評価している.

WDM optical networks represent the direction to the future high capacity wide-area network applications. By creating the optical paths between several nodes in the core networks, logical topology can be created over the physical topology. Reconfiguration of logical topology is necessary for efficient network resource utilization due to the changing traffic. Fast logical topology reconfigurations will be needed to keep up with the traffic pattern changing. At the first place, a fast logical topology design algorithm is necessary to find an optimal or near optimal solution in a short time. In this paper we consider the method of designing logical topology in wavelength routed IP networks when reconfiguration is needed. The exact formulation of the logical topology design problem is usually given as a Mixed Integer Linear Programming, but it grows intractable with increasing size of network. Metaheuristic approaches have been proposed, here we propose a different simulated annealing approach to determine better logical topology within shorter time period, and compare it with other methods.

Multicast communication is a remarkable technology that can substantially decrease network traffic and can save network resources and transmission costs. In multicasting, the transmission speed must be set to the lowest one among the available capacities of links in the multicast tree. This characteristic increases the data transmission time, and accordingly the content could not be delivered efficiently. We have studied an adaptive delivery system to dynamically select multicasting and store- and-forward transferring. This system was proved as an effective system in best-effort networks. This paper proposes the implementation methods of the adaptive content delivery system fit for the Internet considering the current IP multicast routing protocol.

In contents delivery networks, the traffic of large-size contents will cause heavy congestion especially at peak usage hours. One method to cope with this problem is to use a time-dependent multi-class tariff. The price of high priority class is high, and that of low priority class is low. At peak usage hours, the prices will be higher, especially the price of high priority class will be much higher. This leads users to off-peak usage hours, and the traffic of the day will be smoothed. This paper evaluates the effectiveness of this system from the viewpoint of users' utility.

Everything over IP, the concept on which all types of traffic are sent using the Internet Protocol, has been advancing to our lives. Considering the increase of the future traffic and the expansion of its required bandwidth, however, it is inefficient to divide information into many small IP packets. It is thus more desirable for users to choose various transport networks according to the provided services. This paper proposes a new network architecture that supports both the packet-switched network and the circuit-switched network for the next generation networks and evaluates its performance.

In recent years, the communication network environment has changed dramatically. Many carriers are beginning to offer various new communication services, besides conventional services. One of the most attractive services is the differentiated QoS service with multiple priority control. In this paper, assuming the guaranteed bandwidth service with flat rate pricing, the quantitative relation between the guaranteed bandwidth and the user's willingness to pay is analysed. From the questionnaire survey, the relationship between the quality and the user's willingness to pay is obtained. Using the least squares method, the relationship between the quality and the willingness to pay is estimated. Finally, the factors that affect willingness to pay are shown by using the methods of regression analysis and quantification theory I.

Multicast technology is expected to become widely used, since its link sharing capability makes it well suited to large size contents delivery services. In multicast contents delivery services, the waiting time of users will decrease if the contents delivery interval becomes shorter, However, if the contents delivery interval is too short, links will be congested and the waiting time of users will increase. The user's utility on a contents download service depends on the waiting time: the time between his request for the content and the completion of its download. This paper describes how to set the delivery intervals adaptively from the viewpoint of the sun total of users' utility. It is shown that there is an aptimal value of the ocntents delivery interval.

With the rapid growth of the computer network, it is becoming difficult for routers to handle growing amount of routing data. Hierarchical Routing is one of the solutions to this problem. In the hierarchical routing, all nodes are divided into groups in which each node holds simplified data. As the result of simplification, however, it is more difficult for each node to recognize detailed network topology and to select efficient route. Thus, there exists a trade-off relationship between simplification of routing data and its efficiency. In addition, hierarchical grouping of nodes should include consideration of traffic on each link. This paper proposes grouping management method that introduces traffic information into the process of creation of a set of evaluated hierarchical routing groups. The method allows to greatly reduce the size of routing data while keeping low average delay in the network.

Auctions have been proposed and applied to perform resource allocation in reservation-based networks. The methods proposed by these works perceive resources as single items with multiple units and place importance on a limited efficiency inside each node. However, a user does not evaluate each required resource one by one, but evaluates a whole set of required resources. Therefore, the economical efficiency of the overall network cannot be achieved by these methods. To solve this problem, we propose a bandwidth allocation system using Generalized Vickrey Auction. Network resources, which are composed of many links at various bandwidths, are regarded by the proposed method as multiple items with multiple units. We evaluate the efficiency of the proposed method from the viewpoints of the resource utilization and users' utility.

波長ルーチング光ネットワークでは,適切な論理トポロジー設計方法を用いると,遅延を軽減し,ネットワークのスループットを向上させることができる.また,動的に変化するトラヒックに対して,現時点でのトラヒックに大きな混乱を与えないような論理トポロジーの再構築が必要とされている.理想的な論理トポロジーは,多くの場合混合整数線形計画法から得ることができるが,これはネットワークサイズが大きくなるにつれて急速に計算量が増大し,計算不能に陥る.本稿では,ヒューリスティック手法を提案し,論理トポロジーの性能向上と再構築の際のオペレーションの低減を目指している.

Stratospheric platform (SPF) networks can realize efficient multicast service by utilizing multiple address function or inter-links among SPFs. Conventional multicast routing algorithms, however, cannot efficiently operate in SPF networks owing to high cost of multicast trees, much consumption of network resource, or underutilizing the SPF's characteristics. We have developed a new multicast routing algorithm for SPF networks. The algorithm periodically reroutes the multicast trees in each SPF's coverage area. We also evaluated the proposed algorithm. The results show that the proposed algorithm can operate more effeciently than conventional algorithms in SPF networks.

This study introduces Reinforcement Learning (RL) to solve call admission control and routing problem for multi-point connections in networks serving multiple service classes. It presents a model formulating the CAC (Call Admission Control) and Routing as a problem of control, then decomposition is implemented to make the solution manageable. Finally, simulation results demonstrate the prominence of RL-based routing in comparison to MOSPF (Multicast extension to Open Shortest Path First) protocol.

近年,動画像のストリーミング再生が広く行われるようになってきたが,幅輳によりその安定した再生が困難になる場合がある.そこで,ストリーミング配信に優先制御を適用し,受信バッファ内のパケット数の多寡によりパケット優先度を動的に変更する方式を提案している.この方式を用いることにより,網の幅輳状況にかかわらず品質の高い配信が行えることをシミュレーションにより明らかにしている,

In this paper, we consider the method of logical topology reconfiguration in wavelength routed optical networks. The exact formulation of the logical topology problem grows intractable with increasing size of network. Here we proposed a heuristic approach to determine the logical topology reconfiguration with small logical topology changes.

Stratospheric platform (SPF) networks can realize efficient multicast service by utilizing multiple address function or inter-links among SPFs. Conventional multicast routing algorithms, however, cannot efficiently operate in SPF networks owing to high cost of multicast trees, much consumption of network resource, or underutilizing the SPF's characteristics. We have developed a new multicast routing algorithm for SPF networks. The algorithm periodically reroutes the multicast trees in each SPF's coverage area. We also evaluated the proposed algorithm. The results show that the proposed algorithm can operate more effeciently than conventional algorithms in SPF networks.

This study introduces Reinforcement Learning (RL) to solve call admission control and routing problem for multi-point connections in networks serving multiple service classes. It presents a model formulating the CAC (Call Admission Control) and Routing as a problem of control, then decomposition is implemented to make the solution manageable. Finally, simulation results demonstrate the prominence of RL-based routing in comparison to MOSPF (Multicast extension to Open Shortest Path First) protocol.

In video streaming distribution services, the video streams often lose the smoothness because of network congestion. In this paper, an effective algorithm to keep the smoothness is proposed. In this algorithm, the priority of packets is dynamically changed according to the number of packets in a receiving buffer. The computer simulation shows the proposed algorithm is effective.

Multicast technology is expected to become widely used, since its link sharing capability makes it well suited to broadband streaming services. However, multicast services have not become popular. One of the main reasons for the unpopularity of multicast is the lack of a good business model and effective pricing mechanism, as the cost structure of network usage is quite different from that of unicast services. This paper discusses optimal multicast pricing mechanism which maximizes the sender's profits or the networks carrier's profits, when each users behave only to maximize their own benefits. Then, the optimal multicast pricing method is analyzed by computer simulation, which shows the effectiveness of the method.

Multicast technology is expected to become widely used, since its link sharing capability makes it well suited to broadband streaming services. However, multicast services have not become popular. One of the main reasons for the unpopularity of multicast is the lack of a good business model and effective pricing mechanism, as the cost structure of network usage is quite different from that of unicast services. This paper discusses optimal multicast pricing mechanism which maximizes the sender's profits or the networks carrier's profits, when each users behave only to maximize their own benefits. Then, the optimal multicast pricing method is analyzed by computer simulation, which shows the effectiveness of the method.

Multicast technology is expected to become widely used, since its link sharing capability makes it well suited to broadband streaming services. However, multicast services have not become popular. One of the main reasons for the unpopularity of multicast is the lack of a good business model and effective pricing mechanism, as the cost structure of network usage is quite different from that of unicast services. This paper discusses optimal multicast pricing mechanism which maximizes the sender's profits or the networks carrier's profits, when each users behave only to maximize their own benefits. Then, the optimal multicast pricing method is analyzed by computer simulation, which shows the effectiveness of the method.

The demand for the QoS in the Internet, growing heterogeneity of networks, development of more complex routing protocols, various network-based multimedia applications, require knowledge of the bandwidth to be able to operate. Routing devices and protocols are devised with the requirement of prompt forwarding of packets and cannot bear the overhead of including bandwidth information in headers or perform any extra operations with it. Therefore, probing is required to learn of the bottlenecks along a defined path. We defined a dynamic method that probes a path to learn of its bottleneck bandwidth and strives to impose no extra overhead by changing probing parameters, such as packet size and probing frequency. Simulated results are studied and compared with other major works in the field.

The demand for the QoS in the Internet, growing heterogeneity of networks, development of more complex routing protocols, various network-based multimedia applications, require knowledge of the bandwidth to be able to operate. Routing devices and protocols are devised with the requirement of prompt forwarding of packets and cannot bear the overhead of including bandwidth information in headers or perform any extra operations with it. Therefore, probing is required to learn of the bottlenecks along a defined path. We defined a dynamic method that probes a path to learn of its bottleneck bandwidth and strives to impose no extra overhead by changing probing parameters, such as packet size and probing frequency. Simulated results are studied and compared with other major works in the field.

コンテンツ配信において,データ取得までの待ち時間の増加に伴ないユーザ効用は減少する.また,効用の減少の割合はコンテンツの属性に応じて異なると考えられる.そこで,本研究では配信するコンテンツの属性と待ち時間の関係をアンケート調査より明らかにする.そして,効用の変化の主要因となるコンテンツ属性を示す.

In contents delivery service, user utility decreases with the increase in waiting time to get the contents data. The decreasing rate of utility differs according to the contents attribute. In this paper, the relation between the contents attribute and the waiting time to get the contents is clarified by the questionnaire, and the attribute used as the key factor of change of utility is shown.

In contents delivery service, user utility decreases with the increase in waiting time to get the contents data. The decreasing rate of utility differs according to the contents attribute. In this paper, the relation between the contents attribute and the waiting time to get the contents is clarified by the questionnaire, and the attribute used as the key factor of change of utility is shown.

SFQ(Single Flux Quantum) technology has been improved, and various logic circuits using SFQ can be made. Ultra high speed packet switching can be done by using SFQ logic circuits. This paper proposes an architecture of ultra high speed router using SFQ banyan network. This system can deal with variable length IP packets without any additional processing. The performance of banyan network is compared with a simple VOQ (Virtual Output Queuing) packet switch by computer simulation. As a result, the banyan network is more efficient than the simple VOQ.

SFQ(Single Flux Quantum) technology has been improved, and various logic circuits using SFQ can be made. Ultra high speed packet switching can be done by using SFQ logic circuits. This paper proposes an architecture of ultra high speed router using SFQ banyan network. This system can deal with variable length IP packets without any additional processing. The performance of banyan network is compared with a simple VOQ (Virtual Output Queuing) packet switch by computer simulation. As a result, the banyan network is more efficient than the simple VOQ.

Variable kinds of services are expected by a realization of the broadband network. In future, the service which transmit large data is going to increase. Therefore, multicast technology, which ensures the network resource and decrease the transmission cost, is expected. Therefore, pricing is one of the issues for a realization of these services. In multicast, users share the telecommunication links. The question is how the charge of users reflects the fee for the link usage. In this paper, we define multicast cost sharing methods and disucuss the cost shared among users in a multicast-type streaming services.

The computer network is growing rapidly, and is going to be very large. Because of this growth, some problems loomed up. They are too much information for routing, excessive flow of the control traffic, etc. Hierarchical routing is one of the methods to solve the problem of too much routing information. In the hierarchical routing, all nodes are divided into some groups which compose a logical network, and each logical node has simplified information for routing. Simplifying information, however, makes it difficult that each node recognizes the whole network topology and selects the efficient route. Accordingly, the simplification of information and the efficient routing have a trade-off relationship. Therefore, this paper focuses on the grouping methods, and the optimal grouping methods for the hierarchical routing are proposed.

In contents delivery networks (CDN), contents often have large-size data. Such large data causes heavy congestion in the network. Traffic load can be balanced by users' choosing priority classes with accordance of a price table, which shows contents delivery time and price list reflecting network conditions. This paper proposes a traffic load balancing method to show the price table and a scheduling algorithm for completely sending the contents by the promised delivery time.

近い将来,マルチキャスト通信がネットワークトラヒックのかなりの部分を占めると考えられる.現在,インターネット上においてもIPマルチキャストの利用が始まっているが,これはIPと同様にベストエフォート形プロトコルであり,パケット損失の通知,再送,及びパケットの順序保証などの信頼性は保証されていない.IPマルチキャストだけでなく,将来のマルチキャスト通信においても,信頼性の保証は重要な課題となる.そのため,多くのリライアブルマルチキャストプロトコルが提案されている.本稿では,リライアブルマルチキャストにおける再送方式に着目している.これらのプロトコルは,ネットワークトポロジーによって性能が異なると考えられる.そこで,基本的なネットワークトポロジーを組み合わせた階層構成ネットワークにおいて,再送パケットを含めたトラヒック量に基づくネットワーク構成コストによる比較を行うことで,各再送方式を用いた場合の性能評価を行う.

Multicast communication will occupy large proportion of the network traffic in the near future. IP multicast has begun to be used on the present Internet. It is a best-effort-type protocol similar to IP and cannot guarantee the packet loss notification, retransmission, and ordered packet delivery. Guaranteeing such reliabilities is a problem to be solved not only on IP multicast but also on the future multicast network. For that purpose, many reliable multicast protocols have been proposed. Several ways of retransmissions for implementing reliable multicast mechanisms, using unicast or multicast, have already been applied. This paper focuses on the retransmission methods for reliable multicasting. The characteristics of these methods will depend on the network. This paper discusses the suitable retransmission methods on hierarchically structured networks by comparing network costs based on traffic volume including retransmission packets.

Tariff structures for leased line services have been drastically changing, and this aspect will have strong influence on topological policies to design enterprise networks. Otis effective to construct the enterprise networks in consideration of diverse tariff structures in the future. This paper studies the robust enterprise networks, with simulated annealing, against diverse tariff structures considering the future trend. The obtained results show that the best topology at the gravity center in the area of the predicted cost parameters is the most robust topology against those tariff structures.

Recently, the Internet is used not only for conventional applications such as Web and E-mail, but also for new applications such as network meeting and contents delivery. IP multicast is studied for such demands from various points of view. However, IP multicast is not popular now. It is pointed out that there is no incentive to use IP multicast for receivers. In the present IP multicast, unicast packets and multicast packets are treated equally in routers. However, multicast packets should be more weighted than unicast packets, because multicast packets are sent to many receivers. The aim of this study is to improve the characteristics of packet loss and delay in a whole network by performing the control according to the number of receivers of a packet, and to increase the incentive to use IP multicast for receivers. For this control, it is necessary that each router knows the number of receivers of a packet and allocates the bandwidth according to the number of receivers. This paper proposes IP multicast protocol that the router can know the number of receivers, and the bandwidth allocation method using the number of receivers. The simulation results show the good performance of this method.

Recently, IP micoro mobility protocols, such as Hierarchical Mobile IP or CellularIP, have been discussed in the IETF Mobile IP Working Group. However, these protocols are designed for environments where the network is a simple tree structure and only one gateway is deployed. Therefore, when we implement these protocols into real mobile networks which have multiple gateways, we need to add a function which can support a load barancing algorithm for multipule gateways to increase credibility against impediments of the network and reduce delay of packet arrival. In this article, we focus on Cellular IP and propose a load balancing algorithm for robust mobile networks which have multiple gateways. Also, we show simulation results and confirm our proposed method is effective.

Recently, IP micoro mobility protocols, such as Hierarchical Mobile IP or CellularIP, have been discussed in the IETF Mobile IP Working Group. However, these protocols are designed for environments where the network is a simple tree structure and only one gateway is deployed. Therefore, when we implement these protocols into real mobile networks which have multiple gateways, we need to add a function which can support a load barancing algorithm for multipule gateways to increase credibility against impediments of the network and reduce delay of packet arrival. In this article, we focus on Cellular IP and propose a load balancing algorithm for robust mobile networks which have multiple gateways. Also, we show simulation results and confirm our proposed method is effective.

Recently, IP micoro mobility protocols, such as Hierarchical Mobile IP or CellularIP, have been discussed in the IETF Mobile IP Working Group. However, these protocols are designed for environments where the network is a simple tree structure and only one gateway is deployed. Therefore, when we implement these protocols into real mobile networks which have multiple gateways, we need to add a function which can support a load barancing algorithm for multipule gateways to increase credibility against impediments of the network and reduce delay of packet arrival. In this article, we focus on Cellular IP and propose a load balancing algorithm for robust mobile networks which have multiple gateways. Also, we show simulation results and confirm our proposed method is effective.

In a packet network, contentions occur at a switching node, which might cause packet blocking. Previous works on optical packet switching adopted optical delay lines, deflection routing, or wavelength conversion to resolve contention, but they have problems. This paper proposes a new contention resolution scheme that resolves contention by selecting a wavelength of a packet at the edge node, and a new differentiated service scheme that is based on Selective Wavelength Scheme. Their performance level is analyzed by simulation. It is shown that the proposed differentiated selective wavelength scheme is an effective scheme for resolving contention and differentiated service.

光パケットスイッチングでは, スイッチノードにおいて競合が起こり, パケットが廃棄されてしまうという問題がある.遅延線や波長変換装置を用いた方式や迂回ルーチングを行うなどの競合回避方法が従来検討されているが, それらには問題点もある.そこで, 送信ノードで光パケットに対する波長を選択することによって競合回避を行う方式を検討し, 更に波長選択によりサービスクラスごとに品質の差別化を行う差別化波長選択方式を提案する.シミュレーションによる評価により, 差別化波長選択方式が競合回避や品質の差別化に有効であることを示す.

Network resources, which are composed of many links at various bandwidth, can be regarded as multiple items with multiple units. Auctions are known as an effective approach for allocating multiple items with multiple units. Therefore auctions have been proposed and applied to perform contract negotiation and resource allocation in reservation-based networks. However, users' behaviours strongly affect the results of auctions in their efficiency. This paper considers the effectiveness of bandwidth allocation systems by using simultaneous one-round auctions from the viewpoint of users' behaviours.

コンテンツ配信において, 優先度に応じた料金を設定するサービスが考えられる.本稿では, 複数クラスを設けた場合の優先制御における効用と料金に着目した.効用の測度として支払意思額(Willingness to Pay:WTP)を用い, ユーザ全体の効用最大化と通信事業者の収入最大化の最適条件について検討する.また, 複数クラスを設けた場合の各クラスのトラヒックと待ち時間の最適値を示し, 最適な優先クラス数について考察する.

コンテンツ配信において, 優先度に応じた料金を設定するサービスが考えられる.本稿では, 複数クラスを設けた場合の優先制御における効用と料金に着目した.効用の測度として支払意思額(Willingness to Pay:WTP)を用い, ユーザ全体の効用最大化と通信事業者の収入最大化の最適条件について検討する.また, 複数クラスを設けた場合の各クラスのトラヒックと待ち時間の最適値を示し, 最適な優先クラス数について考察する.

For contents delivery systems, the service is considered in which the price depends on each priority class. This paper deals with the multiple priority classes of the contents delivery system from the viewpoint of the utility and pricing. The willingness-to-pay (WTP) is introduced as a measure of utility, which is regarded as the price, and the optimum condition is analyzed to maximize the user's utility as well as the provider's revenue. For the system with multiple priority classes, the optimum condition is given in terms of the traffic load, waiting time for service and price for each priority class. The optimum number of priority classes is also discussed.

Reliable multicast has got much attention as a technology on information delivery systems for the next generation. It guarantees information deliveries without errors on the large-scale multicasting. Several ways of retransmissions for implementing reliable multicast mechanisms, using unicast, multicast, and subcast, have already been proposed. Their characteristics will be changed according to network topologies. This paper focuses on the relation between retransmission method of reliabel multicast and network topology. Namely, this paper discusses the suitable combination of a network topology and a retransmission method by comparing network cost based on traffic volume including retransmission packets.

Rapid technological growth is bringing about price reductions of network components and drastic change of the tariff structures for leased line services. The tariff structure of leased line service has become non-linear cost function for distance and capacity. This paper studies the optimal method to minimize the whole enterprise network cost considering topological connectivity and routing for non-linear cost functions.

In electronic ballot, there are problems of security, such as double votes, alterations of votes, etc. In addition, the organization to add up the votes may leak the result unfairly during its operation. Some methods to prevent them have already been proposed. However, in these methods, voters have to do troublesome operations, or two or more organizations to add up the votes are necessary. In this paper, a new system is proposed. The election administration committee creates the cards that show the PIN numbers, and distributes them to voters randomly and anonymously. The voter submits the card corresponding to the candidate, whom he wants to vote.

In the contents delivery network, its utility depends on the waiting time. This paper deals with the priority control of the contents delivery network from the viewpoint of utility. First, the subjective evaluation experiment of image download is done to obtain the relation between the waiting time and the utility. From the results, the function of utility versus the waiting time is estimated. Then, this function is applied to the optimization of priority control of the image delivery system. This system has two classes, and it is analyzed by using computer simulation. The ratio of high priority packets to low priority packets is changed as a parameter, and the total utility of all users is maximized. It is shown that the total utility of two class system can be higher than that of one class system.

In the contents delivery network, its utility depends on the waiting time. This paper deals with the priority control of the contents delivery network from the viewpoint of utility. First, the subjective evaluation experiment of image download is done to obtain the relation between the waiting time and the utility. From the results, the function of utility versus the waiting time is estimated. Then, this function is applied to the optimization of priority control of the image delivery system. This system has two classes, and it is analyzed by using computer simulation. The ratio of high priority packets to low priority packets is changed as a parameter, and the total utility of all users is maximized. It is shown that the total utility of two class system can be higher than that of one class system.

For data delivery systems with priority classes, an optimum design has been proposed to maximize the utility of the users. As a result, it has been shown that there exist optimum ratios of priority and non-priority traffic and the waiting times [1]. In this paper, we use the willingness to pay(WTP) as a measure of utility, and examine the optimum design to propose the optimum pricing methods. The optimum condition for the service grade (waiting time for service) and pricing for priority and non-priority classes is discussed, considering the request rate which is constant and varying with the utility.

The computer network is growing rapidly, and is going to be very large. Because of this growth, some problems loomed up. They are too much information for routing, excessive flow of the control traffic, etc. Hierarchical routing is one method to deal with the problem of information for routing easily. This paper proposes the optimal grouping method on hierarchical routing.

Multicast service is going to be important in the telecommunication networks. To offer the multicast service to users, its price must be decided. In multicast, users share the telecommunication links. Therefore, users have to share the telecommunication cost. This paper considers the cost share methods, and calculates the probability distribution of the shared cost.

Most companies are constructing enterprise networks between a head office and local branches using leased lines or VPN. The cost structure of leased lines will drastically change in the near future, because of rapid developments in high-speed optical fibers, and the competition among telecommunication operating companies. This paper evaluates the characteristics of optimal enterprise network topologies when the tariff structures change with satisfying some relations.

Most companies are constructing enterprise networks between a head office and local branches using leased lines or VPN. The cost structure of leased lines will drastically change in the near future, because of rapid developments in high-speed optical fibers, and the competition among telecommunication operating companies. This paper evaluates the characteristics of optimal enterprise network topologies when the tariff structures change with satisfying some relations.

複数のリンクから構成され、様々な品質が提供される通信ネットワークにおいてユーザーに配分される通信リソースは、複数種類・複数単位の財であると考えることができる.複数種類・複数単位の財の割り当てを決定する効率的な方式として、一般化ビックレイオークション(GVA:Generalized Vickrey Auction)が知られている.本稿では、専用線の販売・リセールやエンドユーザに対する帯域割当形サービス等を想定し、通信帯域を割り当てる方式としてGVAの適用を考え、シミュレーションによる他の方式との比較評価を通してその適用効果を明確にする.

複数のリンクから構成され, 様々な品質が提供される通信ネットワークにおいてユーザに配分される通信リソースは, 複数種類・複数単位の財であると考えることができる.複数種類・複数単位の財の割り当てを決定する効率的な方式として, 一般化ビックレイオークション(GVA:Generalized Vickrey Auction)が知られている.本稿では, 専用線の販売・リセールやエンドユーザに対する帯域割当形サービス等を想定し, 通信帯域を割り当てる方式としてGVAの適用を考え, シミュレーションによる他の方式との比較評価を通してその適用効果を明確にする.

Network resources, which are composed from many links and qualities of services, can be regarded as multiple items, multiple units. Generalized Vickrey auction is known as an effective auction protocol for allocating multiple items, multiple units. This paper proposes the bandwidth allocation system by using generalized Vickrey auction. The system can be used for selling lines, reselling lines, or allocating bandwidth dynamically to users. The effectiveness of the proposed system is also shown with numerical results using the computer simulation.

コンテンツ配信サービスにおいて, 優先制御を行うことによって, 待ち時間が異なる複数クラスを設けるサービスが考えられる.待ち時間が小さいクラスは高い料金に, 待ち時間が大きいクラスは低い料金になる.本稿では, 配信方式としてストリーミング方式とダウンロード方式の2方式を取り上げ, また, サービスクラス数は簡単のため2クラスとして, コンテンツ配信の効用と料金について検討している.ユーザの効用の測度として, 支払い意思額(Willingness to Pay)を導入した.その結果, ユーザ全体の効用の和を最大にする優先トラヒック・普通トラヒックの割合が存在することを示している.また, それに基いて, 優先クラス, 普通クラスの各々の料金を適切に設定する方法を示している.

For data delivery systems such as video-on-demand service, an optimum design is presented to maximize the utility of the system with priority and ordinary classes. The willingness-to-pay is introduced for a measure of utility, and the optimum design is discussed for the streaming and download schemes. The optimum condition is given in terms of the traffic load, waiting time for service and tariff for the priority and ordinary classes.

The objective of this study is to distribute the bandwidth in a fair manner in the backbone network. Traffic is aggregated between ingress and egress edge nodes, and aggregated traffic is given fair share of bandwidth as much as possible. We first apply several routing schemes when exisiting path information is available. Next, the situation when only the residual capacity is available is considered. In this case, precomputed paths are introduced to restrict the degree of unfairness caused by lack of information.

衛星と地上ネットワークが混在するネットワークでは、衛星の同報性を利用することにより効率的なマルチキャストサービスを実現することが可能となる.しかし, 従来のマルチキャストルーチングアルゴリズムを衛星地上統合ネットワークに適用すると, 計算量, 制御トラヒック量, ツリーコストが大きくなるという問題があった.そこで, 本稿では, ネットワークをいくつかのグループに分け, グループごとに周期的にリルーチングする方式を提案する.また, その特性を評価し, 衛星地上統合ネットワークにおいて提案アルゴリズムが従来アルゴリズムよりも有効に働くことを示す.

Satellite-terrestrial (ST) networks can realize efficient multicast service, in which satellites can interconnect many terrestrial nodes-that is, a multiple address function. Conventional multicast routing algorithms, however, cannot efficiently operate in ST networks because of much computational compelexity, traffic load for controlling networks, and high cost of multicast trees. We have developed a new multicast routing algorithm for ST networks. The algorithm divides all of terrestrial nodes into some groups, and periodically reroutes the multicast trees in each group. We also evaluated the proposed algorithm. The result shows that the proposed algorithm can operate more effeciently than conventional algorithms in ST networks.

The layered coding is one of the remarkable technologies to use network resources efficiently because each receiver can select only a certain number of layers allowed by its communication environment constraints. With conventional dynamic routing algorithms in multicasting the layered streams, many query messages are required in a distributed environment, or an efficient multicast tree cannot be constructed. In this paper, we develop the conventional routing algorithm using predetermined path search, and propose a new effective algorithm for the layered streams. The obtained results show that the proposed algorithm is advantageous compared to conventional ones when nodes are added to or removed from the multicast group during steady-state simulation.

Multicast is required with the rapid spread of the Internet. Particularly, reliable multicast, which gurantees the delivery of information, is required. Local recovery for reliable multicast is the method which can localize the influence of retransmissions by collecting nodes. This paper compares the network cost of several network topologies to find the most appropriate topology for reliable multicast with local recovery.

Rapid technological growth is bringing about price reductions in network components, and the tariff structures of leased line services have been changing drastically. Since most enterprise networks are constructed using leased line services, their effective configurations become very important issues. This paper analyzes the characteristics of enterprise network topologies when the tariff structure changes, and examines the topology that is robust against the diverse tariff structures. The obtained results lead us to important conclusions, which we will use as guidelines when we design enterprise networks.

In the Internet Cache Protocol(ICP), which is a conventional cooperative method used between distributed cache servers, as the number of cache servers increase, the amount of excess data which is created by query messages also increases. Therefore, there is a problem of network traffic increase due to the usage of cache servers which are implemented with ICP. In this paper, to overcome this ploblem, we have developed a new scheme named Cache On the Route(COR), which uses the cache servers that are placed on the route to the original server. We also evaluated the proposed method on excess data amount and delay time to the client by simulation, and show that it is superior to the conventional method.

A new multicast ATM switching network separating unicast and multicast cells, is proposed. This switching network aims to decrease the number of switching elements by separating multicast cells and concentrating them. This switching network is applied to transit switching systems, and a new structure of concentrators and bus interfaces are proposed. As a result, switching elements are reduced, compared to the previous architecture while maintaining the cell loss probability at the same level.

Multicast Protocol Can Realize Contents delivery services such as multimedeia package distribution with low costs. But, in a heterogeneous network, the transmission rate will be limited to the rate of the congested link. In this case, it is better to store the data in a buffer and to send it by unicast than to use a multicast protocol. this paper proposes a contents delivery system which is adaptive to the network states, and shows its perfomance.

従来の遅延時間制約付き動的マルチキャストルーチングアルゴリズムでは, 生成されるツリーにループが発生したり, あるいはループの発生を防ぐためにノードからの新たな参加要求が拒否されてしまうという問題があった.本稿では, あらかじめ各ノードにQoSラベル値を付与し, そのQoSラベル値に基づいて接続すべきノードおよびパスを決定するラベルアルゴリズムを提案する.ラベルアルゴリズムでは, ソースから各ノードまでの遅延時間を制約値以下に抑え, ループ発生や参加要求拒否の発生も抑制することができ, かつ低コストのマルチキャストツリーを構成することができる.さらに, 本稿では, 提案アルゴリズムと従来のアルゴリズムをシミュレーションによって比較評価し, 提案アルゴリズムの有効性を示す.

A new multicast ATM switching network separating unicast and multicast cells, is proposed. This switching network aims to decrease the number of switching elements by separating multicast cells and concentrating them. This switching network is applied to transit switching systems, and a new structure of concentrators and bus interfaces are proposed. As a result, switching elements are reduced, compared to the previous architecture while maintaining the cell loss probability at the same level.

Multicast Protocol Can Realize Contents delivery services such as multimedeia package distribution with low costs. But, in a heterogeneous network, the transmission rate will be limited to the rate of the congested link. In this case, it is better to store the data in a buffer and to send it by unicast than to use a multicast protocol. this paper proposes a contents delivery system which is adaptive to the network states, and shows its perfomance.

Many new multimedia applications involve dynamic multiple participants, have stringent source-to-end delay requirements, and consume large amounts of network resources. A conventional algorithm that allows "two coming paths", where nodes in a multicast tree transmit the several identical data flows, is therefore not practical. We have developed an algorithm for delay-constrained dynamic routing. This algorithm uses a QoS label to prevent the occurrence of "two coming paths", and can construct an efficient multicast tree for any traffic volume. The proposed algorithm was superior to conventional routing algorithms in terms of cost nodes were added to or removed from the multicast group during a steady-state simulation.

一般に, ISP(Internet Service Provider)の網は帯域固定の専用線により構築されているが, ユーザの送出するトラヒック量は時間帯に応じて変化する.本稿では, ISPに対するオークションによる動的帯域割当サービスを提案し, その特性について検討している.第一種電気通信事業者がISP(Internet Service Provider)に対して(あるいは, 一次ISPが二次ISPに対して)オークションにより動的に帯域を割り当てれば, 各ISPはそれぞれのポリシーに従って帯域を確保し, 各ユーザはすべてのISPの接続料金と通信品質に関する情報から自分に最も適したISPを選択することができる.本稿では, 料金と通信品質で定義されるユーザの効用値を指標として, 提案する動的帯域割当サービスと従来の専用線サービスとの比較を行い, 提案するサービスモデルの有効性を示す.

In the leased circuit service, which is used by ISPs(Internet Service Providers), the bandwidth is fixed. However, the traffic changes dynamically. Therefore, we proposed the dynamic bandwidth allocation service using auction method. By using this method, backbone networks can be used effectively as each ISP may secure bandwidth according to their own policies. It is also expected that the Internet users can receive good services, as users may get all information about all ISPs, such as access fee, provided QoS, and select the congenial ISP among all ISPs according to those information. In this paper, we compare the dynamic bandwidth allocation service with leased circuit service by using users' utility to estimate effectiveness of the proposed method.

In the leased circuit service, which is used by ISPs(Internet Service Providers), the bandwidth is fixed. However, the traffic changes dynamically. Therefore, we proposed the dynamic bandwidth allocation service using auction method. By using this method, backbone networks can be used effectively as each ISP may secure bandwidth according to their own policies. It is also expected that the Internet users can receive good services, as users may get all information about all ISPs, such as access fee, provided QoS, and select the congenial ISP among all ISPs according to those information. In this paper, we compare the dynamic bandwidth allocation service with leased circuit service by using users' utility to estimate effectiveness of the proposed method.

The multicast connection is obviously different from the unicast one. The conventional network might degrade the quality of multicast services, and might raise the network construction costs. In this paper, the overlay method is proposed, and evaluated for the multicast network design. The results show that the topology designed by the overlay method, which excludes the complecated calculations, has lower cost than the Delta-Star topology. And, it is also shown that the topology obtained by the overlay method can be used as an initial solution of the heuristic design method, and that it reduces the number of iterations considerably.

The multicast connection is obviously different from the unicast one. The conventional network might degrade the quality of multicast services, and might raise the network construction costs. In this paper, the overlay method is proposed, and evaluated for the multicast network design. The results show that the topology designed by the overlay method, which excludes the complecated calculations, has lower cost than the Delta-Star topology. And, it is also shown that the topology obtained by the overlay method can be used as an initial solution of the heuristic design method, and that it reduces the number of iterations considerably.

With conventional dynamic routing algorithms, many query messages are required in a distributed environment for an efficient multicast routing of any traffic volume. We have developed a dynamic routing algorithm that uses a predetermined path search in which an appropriate multicast path is dynamically constructed by searching only a few nodes. This algorithm can construct an efficient multicast tree for any traffic volume. Simulation has shown that the proposed algorithm is advantageous compared to conventional dynamic routing algorithm when nodes are added to or removed from the multicast group during steady-state simulation.

従来の動的マルチキャストルーチングアルゴリズムでは,問い合わせメッセージ数が多い,あるいはツリーコストを小さく抑えることができないといった問題があった.本稿では,あらかじめ決められた探索経路に従い,接続すべきノードを決定する決め打ち探索を用いた動的マルチキャストルーチングアルゴリズムを提案する.提案アルゴリズムは,あらかじめ決められた探索経路を用いて探索を行うので問い合わせメッセージ数を大幅に少なくでき,かつ低コストのマルチキャストツリーを構成することができる.さらに,本稿では,提案マルチキャストルーチングアルゴリズムと従来のアルゴリズムをシミュレーションによって比較評価し,提案アルゴリズムの有効性を明らかにした.

従来の動的マルチキャストルーチングアルゴリズムでは,問い合わせメッセージ数が多い,あるいはツリーコストを小さく抑えることができないといった問題があった.本稿では,あらかじめ決められた探索経路に従い,接続すべきノードを決定する決め打ち探索を用いた動的マルチキャストルーチングアルゴリズムを提案する.提案アルゴリズムは,あらかじめ決められた探索経路を用いて探索を行うので問い合わせメッセージ数を大幅に少なくでき,かつ低コストのマルチキャストツリーを構成することができる.さらに,本稿では,提案マルチキャストルーチングアルゴリズムと従来のアルゴリズムをシミュレーションによって比較評価し,提案アルゴリズムの有効性を明らかにした.

In the leased Circuit service, which is used by enterprises and ISPs (Internet Service Providers), the bandwidth is fixed. However, the traffic changes dynamically. Therefore, there is strong demand to allocate the bandwidth dynamically. This paper shows the auction method to do it. Auction is one method to allocate resources fairly and efficiently. By using this method, enterprises and ISPs can secure resources as the need changes. Moreover, the network can be used efficiently. In this method, enterprises and ISPs make a bid for bandwidth of the backbone network. Each bid price is decided on the basis of the need to get bandwidth. There are several auction schemes. In this paper, it is described how a network operator can allocate the bandwidth dynamically by using an English auction approach. Then, this method is analyzed from various viewpoints, such as user satisfaction, ISPs' benefit, network operators' benefit, etc.

In the leased Circuit service, which is used by enterprises and ISPs (Internet Service Providers), the bandwidth is fixed. However, the traffic changes dynamically. Therefore, there is strong demand to allocate the bandwidth dynamically. This paper shows the auction method to do it. Auction is one method to allocate resources fairly and efficiently. By using this method, enterprises and ISPs can secure resources as the need changes. Moreover, the network can be used efficiently. In this method, enterprises and ISPs make a bid for bandwidth of the backbone network. Each bid price is decided on the basis of the need to get bandwidth. There are several auction schemes. In this paper, it is described how a network operator can allocate the bandwidth dynamically by using an English auction approach. Then, this method is analyzed from various viewpoints, such as user satisfaction, ISPs' benefit, network operators' benefit, etc.

Distributed Web caching allows multiple clients to quickly access a pool of popular Web pages. Conventional distributed Web caching schemes, e.g., the Internet cache protocol and hash routing, require the sending of many query messages among cache servers and/or impose a large load on the cache servers when they are widely dispersed. To overcome these problems, we propose a hash-based query caching method using both a hash function and a query caching method. This method can find cached objects among several cache servers by using only one query message, enabling the construction of an efficient large-scale distributed Web cache server. Compared to conventional methods, this method reduces cache server overhead and object retrieval latency.

Satellite-terrestrial (ST) networks, in which many nodes are interconnected by both satellite and terrestrial networks, can efficiently support multicast service. An ST network requires a multicast routing algorithm that can select the appropriate satellite and terrestrial routes. We have developed a dynamic routing algorithm for ST networks that selects the route to use according to the multicast group size when a node is added to the group. Simulation showed that the proposed algorithm is advantageous when nodes are added to or removed from the multicast group during steady-state simulation.

Multicast protocol can realize contents delivery services such as multimedeia package distribution with low costs. But, in a heterogeneous network, the transmission rate will be limited to the rate of the congested link. In this case, it is better to store the data in a buffer and to send it by unicast than to use a multicast protocol. This paper proposes a contents delivery system which is adaptive to the network states, and shows perfomance.

We have proposed the Batcher banyan network with cell copy preparation stages, assuming that it would be used for a transit switch. This switching network makes it possible to perform multicast connections with little hardware increase. This paper examines several strategies to prevent the external blocking. We adopt the input buffer with a reservation network as the method. Mainly for the reason of simple control, we propose cells to be reserved after cell duplication. The output of cell copy preparation stages is not uniform and this leads to poor performance. Therefore, we propose to annex a cell distribution network that distributes cells, and examine its effect.

We have proposed the Batcher banyan network with cell copy preparation stages, assuming that it would be used for a transit switch. This switching network makes it possible to perform multicast connections with little hardware increase. This paper examines several strategies to prevent the external blocking. We adopt the input buffer with a reservation network as the method. Mainly for the reason of simple control, we propose cells to be reserved after cell duplication. The output of cell copy preparation stages is not uniform and this leads to poor performance. Therefore, we propose to annex a cell distribution network that distributes cells, and examine its effect.

To implement Connectionless service over ATM exist different alternatives, the most accepted method is using Connectionless servers (CLS) deployed in some nodes of the ATM network. This solution can be generalized if we introduce a specific server called 'Virtual Circuit Manager' (VCM) in order to build different topologies besides the multiple star considered in the CLS approach. To optimize the topology created by the VCM, many algorithms can be used, here is considered the case of Annealing algorithm.

マルチキャスト通信の接続形態は, 従来の1対1通信とは明らかに異なるものであるため, 将来マルチキャスト通信が普及すれば最適な通信網形態が変化すると思われる.筆者らは, マルチキャスト通信に適した通信網構成の検討を行っているが, マルチキャスト通信においては現状のスター構成よりもデルタスター構成と呼ばれるトポロジーが最適であるとの結論を得ている.そこで, 本稿では, マルチキャストトラヒックが増加した場合の各トポロジーの特性変化をコストの観点から調査し, 現状トポロジーからどのような増設を行うことでデルタスター構成へと変化させるべきかを検討した.その結果, 比較的簡単なアルゴリズムを用いて回線増設箇所を決定すれば十分であることがわかった.

Data delivery services such as software distribution, video distribution, and multi-medeia package distribution are expected recently. Since these services are tolerant to the delay, they can be provided at low cost. But, the delay variation is inconvenient for users. This paper proposes a time appointed data-delivery system for the best effort network, and evaluates its performance by computer simulation.

マルチキャスト通信の接続形態は, 従来の1対1通信とは明らかに異なるものであるため, 将来マルチキャスト通信が普及すれば最適な通信網形態が変化すると思われる.筆者らは, マルチキャスト通信に適した通信網構成の検討を行っているが, マルチキャスト通信においては現状のスター構成よりもデルタスター構成と呼ばれるトポロジーが最適であるとの結論を得ている.そこで, 本稿では, マルチキャストトラヒックが増加した場合の各トポロジーの特性変化をコストの観点から調査し, 現状トポロジーからどのような増設を行うことでデルタスター構成へと変化させるべきかを検討した。その結果, 比較的簡単なアルゴリズムを用いて回線増設箇所を決定すれば十分であることがわかった.

Data delivery services such as software distribution, video distribution, and multi-medeia package distribution are expected recently. Since these services are tolerant to the delay, they can be provided at low cost. But, the delay variation is inconvenient for users. This paper proposes a time appointed data-delivery system for the best effort network, and evaluates its performance by computer simulation.

Since the multicast connection form is obviously different from Point-to-Point one, the optimal network topology will be changed if many multicast services are provided. We have been studying the network topology suitable for multicast communication, and it is concluded that Delta-Star is better topology than Star. In this paper, the characteristic of each topology is investigated from the viewpoint with construction cost in case multicast traffic increases. The network extension methods from current topology are also studied. The results show that the simple algorithms work as well as the complicated one.

Data delivery services such as software distribution, video distribution, and multi-medeia package distribution are expected recently. Since these services are tolerant to the delay, they can be provided at low cost. But, the delay variation is inconvenient for users. This paper proposes a time appointed data-delivery system for the best effort network, and evaluates its performance by computer simulation.

The multicast connection form is obviously different from Point-to-Point one. The conventional network might reduce the qualities of Multicast services and might raise the network construction costs. In this paper, the multicast network design methods using heuristic algorithms, Greedy and Simulated Annealing (SA), are proposed from the viewpoint of the network construction cost. The results show that heuristic algorithms are better than the conventional methods that use the fixed link topologies.

The multicast connection form is obviously different from Point-to-Point one. The conventional network might reduce the qualities of Multicast services and might raise the network construction costs. In this paper, the multicast network design methods using heuristic algorithms, Greedy and Simulated Annealing (SA), are proposed from the viewpoint of the network construction cost. The results show that heuristic algorithms are better than the conventional methods that use the fixed link topologies.

An architecture is presented for efficient and reliable delivery of multimedia contents from a primary center (PC) to secondary centers (SCs). Requested contents are delivered from the PC to the SCs through a satellite broadcast channel, or from one SC to another SC through a terrestrial channel. Cycling methods are presented that enable sharing of the contents directory of each SC. Several fundamental models and algorithms are introduced for possible consideration during the planning and design of a contents-delivery system. Simulation has shown that using both satellite broadcast and terrestrial channels for contents delivery is superior in terms of cost to the conventional use of only a satellite network.

WWWやVODのような情報配信サービスを低コストで提供することを目的とした衛生・地上連携ネットワークのアーキテクチャが提案されている.本稿では, ネットワーク条件として新たに提案するホットスポット・キャシング機能とトラヒック条件として各コンテンツの更新される周期に注目し, それらがネットワークコストに与える影響の評価を行う.

WWWやVODのような情報配信サービスを低コストで提供することを目的とした衛星・地上連携ネットワークのアーキテクチャが提案されている.本稿では, ネットワーク条件として新たに提案するホットスポット・キャシング機能とトラヒック条件として各コンテンツの更新される周期に注目し, それらがネットワークコストに与える影響の評価を行う.

Contents-delivery systems using satellite and terrestrial networks for contents-delivery services such as WWW or VOD has been proposed. In this article, we propose "hot-spot caching" for this system, and evaluate the system cost with respect to diverse network and traffic conditions. Especially, we focus on content renewal intervals as traffic conditions.

近年、ソフトウェアやマルチメディアコンテンツを多数のユーザに配送するサービスが期待されている.このようなサービスではマルチキャストを利用することになる.しかし, 特にインターネットのようなネットワークにおいては, マルチキャストを高信頼に行うために, 最も混雑しているリンクの速度に合わせてデータを送信する必要があるので非効率的である.そこで, ネットワーク内の主要ノードにバッファを装備した配送サーバを配置し, マルチキャストツリー全体を, 配送サーバをソースとするサブツリーに分割することで, 効率的な配送を可能とする手法を提案する.また, シミュレーションによりその特性を評価する.

We have proposed the Batcher banyan network with cell copy preparation stages for multicast switching. Hardware volume of this switch is little larger than that of Batcher banyan switching network. And the blocking rate is very even if multicast is done. But, if the number of copies is large, the blocking rate is not small. This paper proposes a new copy cell split method for Batcher banyan network with cell copy preparation stages. The blocking rate decreases significantly by using this method.

B-ISDNが実現されると、放送をはじめとする各種マルチキャストサービスがトラヒックのかなりの部分を占めると考えられる。マルチキャストサービスでは、伝送路上のトラヒックを減らすために、1対多接続が用いられるが、これは従来の1対1接続とは大きく異なる。従って、マルチキャストサービスと1対1通信サービスが混在する場合の網構成を検討する必要がある。一方、1対1通信網の信頼性を重視した場合に有効なトポロジーとして、デルタスター構成が提案されているが、本構成はマルチキャストに有利なリング構成と1対1通信に有利なスター構成の両方を合せたような形状になっているため、両サービスが混在する場合に有効なトポロジーであると考えられる。本稿では、比較的小さいエリアにおいて、一つのトポロジーを用いてマルチキャスト通信網を構築した場合の最適形態についての検討を行った。その結果、多くの場合においてデルタスター構成が最適であることがわかった。

With the realization of B-ISDN, various kinds of multicast services will be introduced into network traffic. Multipoint connection paths, used to transmit multicast services, are quite different from current point-to-point connection paths. Therefore, new network topologies must be designed. On the other hand, Delta-Star has been propoesed as the effective topology in case that point-to-point network reliability is considered. Delta-Star may be suitable for B-ISDN network since this has both Star topology, suitable for point-to-point connections, and Ring, suitable for the multipoint connections. In this paper, the multicast network in local area is optimized by using one of several topologies. The results show that Delta-Star topology is optimal in many cases.

Multicast services will occupy a large part of the B-ISDN traffic. But, even if one multicast service is watched by many subscribers, it occupies only one channel on the trunkline. Namely, the multicast traffic is not dominant on the trunkline. Therefore, it is a waste to use non-blocking multicast ATM Swithing networks for transit switching systems. Because, the hardware of a non-blocking multicast ATM switching network is large. This paper proposes to apply Batcher banyan network, which is point-to-point non-blocking, to multicast transit switching systems. The blocking rate decreases substantially by placing a cell copy preparation stage in front of Batcher network.

With a realization of B-ISDN, broadcast services will occupy a large part of the B-ISDN traffic, and network resources must be saved using efficient routing theory. Generally, routing is not needed in the hierarchical network. However, routing is necessary in ATM network. Because it has many Virtual Paths and it logically becomes non-hierarchical. In this paper, dynamic routing is applied to the ATM network which is phisically hierarchical. The result shows that it is enough to use only static routing and unnecessary to use dynamic routing which requires complex controls, since the influence of the physically hierarchical network comes out strongly.

Multicast services will occupy a large part of the B-ISDN traffic. But, even if one multicast service is watched by many subscribers, it occupies only one channel on the trunkline. Namely, the multicast traffic is not dominant on the trunkline. Therefore, it is a waste to use non-blocking multicast ATM Swithing networks for transit switching systems. Because, the hardware of a non-blocking multicast ATM switching network is large. This paper proposes to apply Batcher banyan network, which is point-to-point non-blocking, to multicast transit switching systems. The blocking rate decreases substantially by placing a cell copy preparation stage in front of Batcher network.

With a realization of B-ISDN, broadcast services will occupy a large part of the B-ISDN traffic, and network resources must be saved using efficient routing theory. Generally, routing is not needed in the hierarchical network. However, routing is necessary in ATM network. Because it has many Virtual Paths and it logically becomes non-hierarchical. In this paper, dynamic routing is applied to the ATM network which is phisically hierarchical. The result shows that it is enough to use only static routing and unnecessary to use dynamic routing which requires complex controls, since the influence of the physically hierarchical network comes out strongly.

Broadcast services will occupy a large part of the B-ISDN traffic. But, even if one broadcast service is watched by many subscribers, it occupies only one channel on the trunkline. Namely, the broadcast traffic is not dominant on the trunkline. Therefore, it is a waste to use non-blocking broadcast ATM Swithing networks for transit switching systems. Because, the hardware of a non-blocking broadcast ATM switching network is large. This paper proposes to apply Batcher banyan network, which is point-to-point non-blocking, to broadcast transit switching systems. The blocking rate decreases substantially by placeing a simple network in front of Batcher network.

Broadcast services will occupy a large part of the B-ISDN traffic. But, even if one broadcast service is watched by many subscribers, it occupies only one channel on the trunkline. Namely, the broadcast traffic is not dominant on the trunkline. Therefore, it is a waste to use non-blocking broadcast ATM Swithing networks for transit switching systems. Because, the hardware of a non-blocking broadcast ATM switching network is large. This paper proposes to apply Batcher banyan network, which is point-to-point non-blocking, to broadcast transit switching systems. The blocking rate decreases substantially by placeing a simple network in front of Batcher network.

Broadcast services will occupy a large part of the B-ISDN traffic. In the broadcast distribution services, calls originate and terminate at any time. Therefore, the dynamic broadcast routing is needed. Since the ATM network has many virtual paths, the dynamic broadcast routing will be more complicated than that of the conventional network. In this paper, the conventional dynamic broadcast routing methods are evaluated. Then, a new dynamic broadcast routing method which is fit for ATM network, is proposed.

We proposed a new broadcast ATM switching fabric, which separates the routes of unicast cells and broadcast, cells, and showed the possibility that the number of switch elements used in it was smaller than that of fully interconnected fabrics such as the Knockout Switch. Though the possibility depends on the number of output lines from the concentrator, the probability of losing cells becomes great if the number is too small. So, in this paper, we study the characteristic of the concentrator and cell loss probability, and analize how much smaller the number of switch elements in our fabric can be than that of the fully interconnected fabric.

With the realization of B-ISDN, broadcast services will be a large part of the network traffic. In these services, the same information is distributed to lots of viewers by multicast connection. This service form is quite different from those of current point-to-point services. In this service, current network topologies may lower the quality of services or may increase the construction costs. In this paper, the hierarchical network topology is optimized under the assumption that the broadcast services will account for a large part of the network traffic. The difference between the destination distribution for point-to-point services and that for broadcast services is considered, and the construction costs for various parameters are calculated.

With the realization of B-ISDN, broadcast services will account for a large part of the network traffic. Since call loss probability must be very small in broadcast services, congestion control is very important in broadcast routing. In this paper, some methods of congestion control are proposed, and their characteristics are shown by simulation. As a result, it is shown that we should select a control method according to the traffic condition.

With the advance of B-ISDN, the demand for broadcast services will be a large part of the network traffic. So, various types of broadcast switching fabric have been proposed. If such proposed fabrics are adapted to the local switching systems, call control will be very complex or the number of switch elements will be very large. In this paper, we propose a new broadcast ATM switching fabric, which can solve such problems.

With the advance of B-ISDN the demand for broadcast services will be a large part of the network traffic. So, a large number of broadcast switching fabrics has been proposed. In the broadcast switching systems, particularly in the local switching systems, the process of rewriting cell header conversion tables is much more complex than point-to-point switching systems. In this paper, we make the operation clear and examine the quantity of call control in the broadcast ATM switching system.

With the realization of B-ISDN, broadcast services will be a large part of the network traffic. In these services, the same information is distributed to lots of viewers by multicast connection. This service form is quite different from those of current point-to-point services. In this service, current network topologies may lower the quality of services or may increase the construction costs. In this paper, the hierarchical network topologies for broadcast connection are analyzed on the assumption that broadcast services will account. for a large part of the network traffic. The construction costs for various values of parameters are calculated.

With the realization of B-ISDN, broadcast services will be a large part of the network traffic. In the broadcast distribution service, the same information is distributed to lots of viewers (multicast connection). This service form is quite different from those of current point-to-point services. In this service, current network topologies may lower the quality of services or may increase the construction costs. In this paper, the network topologies for multicast connection are analyzed on the assumption that broadcast services will account for a large part of the network traffic. The construction costs for various values of parameters are calculated.

The dynamic characteristics of heat exchangers are described by partial differential equations .But the expressions are rather complicated and not well suited for practical use, so it is necessary to develop a simple lumped-parametered model to estimate the dynamic behavior of systems with heat exchangers and also to design optimal control systems. The aim of this paper is to present a relatively simple dynamic model for steam liquid heat exchangers via the method of weighted residuals (MWR). A 2nd-order MWR model is derived with a good approximation, which is represented by the three parameters, i.e. k_p which depends upon the steam and liquid temperature in steady state, τ_p which is the residence time of the liquid, and k_s which is the ratio between the shell-side time constant and τ_p. The results of an experimental study are also represented to validate the analytical results.