Recently deep learning based image compression has made rapid advances with promising results based on objective quality metrics. However, a rigorous subjective quality evaluation on such compression schemes have rarely been reported. This paper aims at perceptual quality studies on learned compression. First, we build a general learned compression approach, and optimize the model. In total six compression algorithms are considered for this study. Then, we perform subjective quality tests in a controlled environment using high-resolution images. Results demonstrate learned compression optimized by MS-SSIM yields competitive results that approach the efficiency of state-of-the-art compression. The results obtained can provide a useful benchmark for future developments in learned image compression.

In this paper, we provide a detailed description on our approach designed for<br />
CVPR 2019 Workshop and Challenge on Learned Image Compression (CLIC). Our<br />
approach mainly consists of two proposals, i.e. deep residual learning for<br />
image compression and sub-pixel convolution as up-sampling operations.<br />
Experimental results have indicated that our approaches, Kattolab, Kattolabv2<br />
and KattolabSSIM, achieve 0.972 in MS-SSIM at the rate constraint of 0.15bpp<br />
with moderate complexity during the validation phase.

Compression has been an important research topic for many decades, to produce a significant impact on data transmission and storage. Recent advances have shown a great potential of learning based image and video compression. Inspired from related works, in this paper, we present an image compression architecture using a convolutional autoencoder, and then generalize image compression to video compression, by adding an interpolation loop into both encoder and decoder sides. Our basic idea is to realize spatial-temporal energy compaction in learning image and video compression. Thereby, we propose to add a spatial energy compaction-based penalty into loss function, to achieve higher image compression performance. Furthermore, based on temporal energy distribution, we propose to select the number of frames in one interpolation loop, adapting to the motion characteristics of video contents. Experimental results demonstrate that our proposed image compression outperforms the latest image compression standard with MS-SSIM quality metric, and provides higher performance compared with state-of-the-art learning compression methods at high bit rates, which benefits from our spatial energy compaction approach. Meanwhile, our proposed video compression approach with temporal energy compaction can significantly outperform MPEG-4, and is competitive with commonly used H.264. Both our image and video compression can produce more visually pleasant results than traditional standards.

This paper introduces recent trends in video streaming and four methods proposed by the authors for video streaming. Video traffic dominates the Internet as seen in current trends, and new visual contents such as UHD and 360-degree movies are being delivered. MPEG-DASH has become popular for adaptive video streaming, and machine learning techniques are being introduced in several parts of video streaming. Along with these research trends, the authors also tried four methods: route navigation, throughput prediction, image quality assessment, and perceptual video streaming. These methods contribute to improving QoS/QoE performance and reducing power consumption and storage size.

To ensure good quality of experience for user when transmitting video content, throughput prediction can contribute to the selection of proper bitrate. In this paper, we propose a throughput prediction method with recurrent neural network (RNN) model. Experiments are conducted to evaluate the methods, and the results indicate that proposed method can decrease the prediction error by a maximum of 29.39% compared with traditional methods.

Image quality assessment (IQA) is an inherent research topic in image processing field for several decades. Recently, machine learning has achieved success in many multimedia tasks and can be applied in IQA. Especially, screen content images (SCIs) is greatly increasing in various applications, but the characteristics of SCIs makes it difficult to directly apply general IQA methods to predict qualities. In this paper, we propose a fast no-reference SCIs quality prediction method. First, we use the convolutional neural networks (CNNs) to predict the quality scores of each patch. Second, we present a SCIs-oriented quality aggregation algorithm for acceleration. Experimental results demonstrate that our method can achieve the high accuracy (0.957) with subjective quality scores, outperforming existing methods. Moreover, our method is computationally appealing, achieving flexible complexity performance by selecting different groups of patches.

The popularity of high dynamic range (HDR) makes the inverse tone mapping become an important technique for HDR display. In this paper, we propose a convolutional neural network (CNN) based inverse tone mapping method to generate a high-quality HDR image from one single standard dynamic range (SDR) image. First, we present a CNN design with a three-channel input, which considers both luminance and chrominance. Second, we propose to use overlapped inputs to remove the boundary artifacts, caused by zero padding in CNN. Experimental results demonstrate the high quality of our generated HDR images compared to the ground truth and conventional inverse tone mapping methods.

In order to recognize the transportation modes without any additional sensor devices, we propose a recognition method by using communication quality factors. In the proposed method, instead of Global Positioning System (GPS) and accelerometer sensors, we collect mobile TCP throughputs, Received Signal Strength Indicators (RSSIs), and cellular base station IDs (Cell IDs) through in-line network measurement when the user enjoys mobile services, such as video streaming service. In accuracy evaluations, we conduct two different field experiments to collect the data in five typical transportation modes (static, walking, riding a bicycle, a bus and a train,) and then construct the classifiers by applying Support Vector Machine (SVM), k-Nearest Neighbor (k-NN) and Random Forest (RF). Results conclude that these transportation modes can be recognized by using communication quality factors with high accuracy as well as the use of accelerometer sensors.

In this paper, to sustain a high-quality 24-hour video surveillance (i.e., high reliability) and reduce redundant video traffic volume (i.e., network friendliness), we propose an edge-centric video surveillance system that provides flexible adaptive control of the image enhancement process and video quality based on an event-driven adaptation. In the proposed system, the video bitrate is adaptively controlled according to the contrast of captured videos and conditions in a monitored area (e.g., 'normal', 'caution', and 'alert'). To confirm the system performance, we evaluate objective image quality, accuracy of human detection and video traffic volume generated by the proposed system. Evaluations conclude that the system can reduce the video traffic while sustaining high-quality visibility.

Adaptive rate video streaming on a novel Internet architecture, called Named Data Networking (NDN), appears to be a promising technology. However, given the source of content on NDN is unknown to the consumer, rate adaptation becomes inefficient. Moreover, the traffic at some nodes holding popular contents still can be congested. In this paper, we utilize the Internet of Things (IoT) device for assisting content delivery at an NDN router that has low bandwidth. The location of IoT devices installation is determined by a metric calculated from link bandwidth, delay, and utilization. We performed the experiments using DASH-NDN-JS on the setup network. The results demonstrate that our proposal could support NDN router to deliver video content to additional consumers efficiently. Thus, the network scalability can be improved with a reasonable cost.

Named Data Networking (NDN) has been recently introduced as a new future network architecture. The end-to-end throughput estimation for adaptive bit-rate video streaming on NDN is one of the most challenging topics. Specifically, the end-to-end throughput estimation on NDN appears to be unreliable, since the provider of content is unknown to the consumer. Moreover, partial caching on NDN router's Content Store could temporarily lead to packet loss due to throughput overestimation. In this paper, we present an active Interest adaptation scheme which operates by proactively estimating the throughput in a hop-by-hop fashion. The consumer node is then assisted with the most recent available end-to-end bandwidth. Therefore, the video player can promptly adapt to the change of network condition. The implementation evaluation using NDN-JS and DASH-JS on the setup network demonstrate that our proposed solution provides better average stream bit-rate and consumes less network bandwidth than the traditional system.

Throughput prediction is essential for ensuring high quality of service for video streaming transmissions. However, current methods are incapable of accurately predicting throughput in mobile networks, especially for moving user scenarios. Therefore, we propose a TCP throughput prediction method TRUST using machine learning for mobile networks. TRUST has two stages: user movement pattern identification and throughput prediction. In the prediction stage, the long short-term memory (LSTM) model is employed for TCP throughput prediction. TRUST takes all the communication quality factors, sensor data and scenario information into consideration. Field experiments are conducted to evaluate TRUST in various scenarios. The results indicate that TRUST can predict future throughput with higher accuracy than the conventional methods, which decreases the throughput prediction error by maximum 44% under the moving bus scenario.

Throughput prediction contributes a lot to adaptive bitrate control, adjusting the quality of video streaming accordingly to offer smooth media transmission and save energy at the same time. To solve the problem of throughput prediction for real time communication, this paper puts forward a new history-based throughput prediction method applying Hidden Markov Model in mobile networks. The main purpose of this method is to predict future throughput for real time communication in mobile network. Our novel approach utilizes Hidden Markov Model (HMM) with Gaussian Mixture Model (GMM) to deal with history time series of throughput and judge fluctuation factor with total variance when predicting future throughput. By conducting experiments with the new methodology, we compare the accuracy of the proposed method with three other conventional prediction methods. Results show our proposed method could identify data fluctuation effectively and predict future 100s throughput with high accuracy in various situations.

With increase of mobile wireless LAN systems, frequency contamination by the overlapping basic service set (OBSS) becomes a critical issue. In this paper, we focus on HCF controlled channel access (HCCA) to alleviate the OBSS problem. HCCA considers single BSS (SBSS) environment only and suppresses traffic of co-existing DCF (distributed coordination function) based WiFi access points (APs). We propose two methods that utilize capture effects to improve the coexistence capability with DCF networks, that we call DCF friendliness. The first method adjusts transmission timing in HCCA WLAN by applying inter-AP coordination. The second method enables simultaneous communication with surrounding DCF WLANs by changing the frame interval of HCCA to DIFS+1 SlotTime. Simulations show that both the proposed method can achieve higher throughput and better DCF friendliness.

In this paper, we propose a method to improve H.265/HEVC encoding performance for 8K UHDTV moving pictures by detecting amount or complexity of object motions. In 8K video, which has high spatial resolution, motion has a big influence on encoding efficiency. The proposed method estimates motion complexity by external process, and selects an optimal prediction mode and search ranges of motion vectors. By this method, we achieved highly efficient and low computation encoding.

In order to provide high-quality and highly reliable video delivery services for mobile users, especially train passengers, we propose a proactive content caching scheme that uses transportation systems. In our system, we place content servers with cache capability [e.g., content centric networking/named data networking (CCN/NDN)] in every train and station. Video segments encapsulated by MPEG-Dynamic Adaptive Streaming over HTTP (MPEG-DASH) are distributed and pre-cached by the station servers before the trains arrive at the stations. The trains receive content via high-speed wireless transport, such as wireless LANs or millimeter waves, when they stop at the stations. We developed prototype systems based on hypertext transfer protocol and CCN/NDN protocol, evaluate their performance through two field experiments that uses actual trains, and compare with traditional video streaming over cellular networks. Such evaluations indicate that our system can achieve high-quality video delivery without interruption for up to 50 users simultaneously.

In this paper, we introduce a content offload scheme utilizing transportation systems to reduce uploading time. In our proposal, when a mobile user experiences network congestion or communication failure, his/her content data are not uploaded to a content server directly, but are stored by a server located on the transportation systems temporarily and forwarded to the content server after the transportation systems move to other areas where communication quality is high. We employ an analytical model for this scheme and evaluate its performance by using computer simulations and model based analysis. Evaluations conclude that our scheme can drastically reduce uploading time rather than a cellular congestion case as the size if upload data gets larger.

In video coding, image quality degradation is inevitable due to quantization steps in the encoding process. We had proposed a blind type (no-reference type) PSNR estimation method which uses SVM and natural image statistics. In this paper, we apply a spatio-temporal segmentation method into the conventional method. As a result, we reveal that our method provides more stable results even when the conventional method fails, and can reduce an estimated error of PSNR.

In this paper, we propose a method to control the way of encoding by H.265/HEVC for 8K UHDTV moving pictures. In 8K video, with high resolution, amount of motion is very large. The proposed method estimates the amount of motion by the external process, and sets the search range of motion vectors according to estimated motion amount. At this time, the method expands search ranges per GOP interval to exploit vector interpolation capability by HM. It also selects intra only prediction for movies which have random motion. By this method, we achieve high efficient and low computation encoding.

Information Centric Networking (ICN) is a new networking paradigm in which the network provides users with named content, instead of communication channels between hosts. However, many issues, such as naming, routing, resource control, and security, still need to be resolved before it can be realized practically. Further, the energy efficiency of ICNs has not been sufficiently considered. In this study, we evaluate the energy consumption of an ICN network by first measuring the power consumption of actual PCs/smartphones and constructing an energy consumption model. Then, by using the model and the simulator ndnSIM, we calculate the energy consumed by the network. We also evaluate the energy consumed for video delivery, with a view toward power-saving video delivery.

In order to tackle recent mobile traffic explosion, content offloading using Wifi access points from cellular network is deployed but the Wifi system itself is sometimes heavily congested. In this paper, we propose a content offloading system using transportation systems such as a train. In addition, we aim to incorporate CCN (Content Centric Networking) into our system to exploit its in-network caching capability. We compare three protocol designs, pure IP, IP+CCN hybrid and pure CCN, and show the efficiency of our system using a network simulator.

In recent years, researches of underwater sensor networks have continued to investigate environment and resources of the sea. Acoustic waves are used instead of radio waves for wireless communication in underwater. However, dedicated hardware is very expensive, experiments on the sea are very time-consuming, and huge water spaces are necessary to study the underwater acoustic communication. In this paper, we present a cheap and tractable software-defined acoustic communication platform running on PCs using MATLAB, and evaluate its characteristics in a variety of communication methods by changing modulation schemes, error correction codes, transmission power and frequency by using commercial speaker and microphone devices. Our current implementation achieves data rate of up to 4.5 Kbps.

A system is proposed that uses content-centric networking to deliver content to mobile users. It uses sensor data obtained from the device they are using to identify their means of transport. The content is then delivered in a manner that best matches that means. The means of transport is identified using a combination of features: average magnetic quantity, average value and standard deviation of acceleration, vehicle speed, and maximum power spectrum of acceleration.

An accurate (reliable) motion estimation (ME) algorithm using block matching for sequences including spatially periodic patterns is proposed and evaluated in comparison to conventional algorithms. Mismatch between accurate motion vectors (MVs) and MVs obtained by block matching (BM) which minimizes BM distortion is evaluated. BM distortion caused by MV sampling error (DVSE), which causes ME errors, is discovered. A new ME algorithm which uses the maximum value of DVSE (maxDVSE) is proposed. As for the proposed algorithm, maxDVSE is calculated for each block as a threshold of accurate vector candidates, and all candidate vectors for each block in a frame are stored. Accurate vectors are selected from the candidate vectors by finding so-called "anchor vectors" (namely, vectors at a unique point on spatially periodic patterns). The MVs are selected in a process starting from one block and propagated to neighboring blocks. Simulations are performed and the results show that average precision of the proposed ME algorithm is 98.9%, which is superior to that of other conventional algorithms.

Proactive content caching scheme utilizing transportation systems, especially on trains, was proposed in order to provide a robust content delivery with efficient wireless resource usage. This system requires content servers with NDN capability to be placed on every station and trains. The mechanism is to pre-cache the contents that users request to the station server before the train arrives, and the train server caches the content during its stoppage time. With this mechanism, users are able to have a continuous playback of videos while riding on trains. In this paper, we have proposed a browser-based implementation, called DASH-NDN-JS, for this proactive content caching scheme. We evaluate this scheme, and experiment with multiple users to see how it will affect the video quality each user will achieve and the bandwidth consumption between the connections. Our evaluations conclude that the increase of users lowers the video quality, but avoids congestion depending on what video content each user will want to request.

Block-based Motion Estimation (ME) has been widely used in various video coding standards to remove temporal redundancy. However, this ME has limitation that it can only compensate for a parallel translation. Various methods have been proposed for other motions such as zooming. In recent years, 8K UHDTV (7,680 x 4,320 pixels) has been developed. Since 8K has large motion by zooming that is difficult to be predicted by block matching, it is important to improve zoom motion estimation. In this paper, to handle zooming in 8K video sequences, we propose a method for improving the picture quality by global zoom estimation based on motion vector analysis extracted by block matching.

Providing robust content delivery service, such as Video on Demand, with efficient wireless resource usage is important for mobile users. To achieve this, we have proposed a proactive content caching scheme utilizing transportation systems (e.g., train). Because our previous prototype is immature and has a scalability issue, in this paper, we introduce an efficient transmission mechanism of Interest messages to improve throughput efficiency in mobile NDN, develop two prototypes based on NDN and HTTP, and then evaluate prototype performances by performing larger-scale field experiments on actual commercial railroad line. Evaluations conclude that our system can achieve high efficient content transfer and fully utilize wireless network bandwidth. As a result, our system can provide high-quality video streaming for up to 50 users simultaneously.

© Springer International Publishing Switzerland 2015. This paper presents a live version music identification system by modifying the conventional cover song identification system. The proposed system includes two stages: a live version identification phase and an audio scenedetection phase. We improve the accuracy of the system by weighting similarity scores in the live version identification phase and discriminating scenes by using RMS, pulse clarity and similarity scores. Results show that the proposed method performs better than the previous method. The final algorithm achieves 70% accuracy on average.

Providing robust video streaming along with efficient wireless resource usage is necessary for mobile users, especially on subway, and mobile carriers. To achieve this, we propose a highly-reliable buffer strategy based on long-term throughput prediction. Our approach has two elements which are called "long-term throughput prediction" and "guaranteed playout buffer filling mechanism." To avoid any video freeze due to network quality degradation, our approach calculates the optimal amount of playout buffer and schedules video download timing in a theoretical manner. We evaluate its performance via experiments in real environment. Evaluations conclude that our approach can provide highly-reliable video streaming and also achieve to reduce the average playout buffer size on the client.

In wireless networks, it is important to realize energy-efficient video delivery. To do this, we introduce energy-efficient video streaming over named data networking (NDN). In our proposed approach, we focus on two areas, namely, to improve the throughput performance by Interest aggregation, and to reduce the overhead energy using playout buffer-size control. We evaluate the power savings realized by our method using a hardware power-measurement tool. Our results show that Interest aggregation can realize an approximately 50% reduction in energy compared to conventional NDN implementation, and a large playout buffer size can reduce the energy by approximately 37% compared to the case of a small playout buffer size.

Content Centric Networking (CCN) has been emerged as a new networking paradigm. In CCN, adaptive video streaming is one of the most promising technologies. However, rate adaptation and outbound face selection are the primary concern since a node cannot reliably estimate the end-to-end throughput. In this paper, we propose a strategy for outbound face selection considering both response time and bit-rate of lost packets. These two aspects facilitate the node to choose which outbound face offers the best QoE. In a case where bit-rate of lost packets is greater than a half of transmitted bit-rate, the node will re-select outbound faces to avoid intermittent streaming. The experiments performed using CCNx demonstrate that our approach provides better average stream bit-rate than the earlier response time based method.

In this paper, we evaluate performances of 4K/2K video streaming using DASH-JS over 2.4/5GHz WiFi networks. We confirmed that fluctuation of reception quality was mainly caused by the minimum representation rate of MPEG-DASH contents and overlapped ON cycles for packet transmission.

Effective use of various wireless network interfaces is expected. In this paper, we evaluate Quality of Service characteristics of mobile 4K video using a radio communication quality map. We collect radio communication quality (e.g. throughput) by smartphones and visualize the quality in a heat map. Using this map, we select two routes that are expected to bring the highest and the lowest communication quality. We then evaluate QoS characteristic of 4K video streaming using MPEG-DASH on these two routes. The result shows that a good route user can achieve 1.5 times higher communication quality than a bad route user.

In this paper, we propose a method to control H.265/HEVC encoding for 8K UHDTV moving pictures by detecting amount or complexity of object motions. In 8K video, which has very high spatial resolution, motion has a big influence on encoding efficiency and processing time. The proposed method estimates motion features by external process which uses local feature points matching between two frames, selects an optimal prediction mode and determines search ranges of motion vectors. Experiments show we can detect motion complexity of 8K movies by using local feature matching between frames and we can select optimal configurations of encoding. By our method, we achieved highly efficient and low computation encoding.

In this paper, we propose a "comfort route" navigation system with proactive content delivery and an offloading scheme for mobile users. To search an optimal path for highly reliable video streaming, we apply a delivery scheduling algorithm, which has been proposed for proactive content delivery scheme, to a path-finding problem. In preliminary evaluation, the result concludes that our approach can drastically reduce transmission time rather than a cellular usage case. This is because the mobile user can communicate with only high-speed hot spots. So, we expect that our approach could also save user's mobile battery life.

In this paper, we evaluate Quality of Service characteristics of MPEG-DASH using 4K video over various wireless networks. We also evaluate the effects of various environments, such as the number of users who share the same wireless channel or bandwidth and users' behaviors (e.g. still, walking, and moving on subway), over Long Term Evolution (LTE), 2.4 GHz and 5 GHz Wi-Fi networks. The results conclude that, in case of LTE and 5 GHz Wi-Fi networks, mobile users enjoy high quality 4K video without any video playback interruption in any situations, even on subway. However, in case of 2.4 GHz Wi-Fi networks, mobile users are forced to experience very low quality video with long video playback freeze. This is because network quality of 2.4GHz Wi-Fi network is quite low due to interference waves from other access points of which wireless channels are severely overlapped.

In the future, network traffic will be increasing according to demand of big traffic of moving picture on the mobile communication environment. For processing large traffic, virtualized network based on SDN (Software defined Network) has been provided for purpose of cloud networking as CDN (Contents Delivery Network). On the other hand, the ICN (Information Centric Network) has been proposing as IP independent network which isn't based on IP address. And it has been discussing about advantages compared to the IP network. This paper shows the experimental results of Proactive Content Caching and Delivery Scheme utilizing Transportation Systems. This research has been supported by SCOPE organized by Ministry of Internal Affairs and Communications.

In this paper, we propose an adaptive buffer control technique for robust video streaming. Our approach has two elements which are called "long-term throughput prediction" and "guaranteed playout buffer filling." Since network quality degradation can be detected by using long-term throughput prediction, playout buffer also can be adaptively and proactively controlled. To avoid any video playback interruption due to the network quality degradation, our approach calculates the optimal amount of playout buffer and schedules video download timing in a theoretical manner. We evaluate its performance in real environment and then validate the effectiveness of our proposal.

We evaluated various characteristics of 4K video streaming using DASH-JS over LTE, public Wi-Fi services and WiMAX. We did experiments while stopping, walking, and riding the train. We found that a mobile user could enjoy smooth video streaming by using these networks even 4K video.

We propose proactive content caching scheme utilizing transportation systems for robust video streaming with efficient wireless resources usage. In this paper, we develop a prototype system and evaluate its performances by using actual train vehicles and computer simulations. Evaluations conclude that our system can achieve higher-speed and highly-reliable video streaming rather than traditional CDN scenarios using cellular networks.

We propose proactive content caching scheme utilizing transportation systems for robust video streaming with efficient wireless resources usage. In this paper, we develop a prototype system and evaluate its performances by using actual train vehicles and computer simulations. Evaluations conclude that our system can achieve higher-speed and highly-reliable video streaming rather than traditional CDN scenarios using cellular networks.

We propose proactive content caching scheme utilizing transportation systems for robust video streaming with efficient wireless resources usage. In this paper, we develop a prototype system and evaluate its performances by using actual train vehicles and computer simulations. Evaluations conclude that our system can achieve higher-speed and highly-reliable video streaming rather than traditional CDN scenarios using cellular networks.

In recent years, opportunities to watch video contents while moving have increased and it is predicted that video contents will cover approximately 70% of the mobile traffic in 2017. We have tried power saving transmission of a mobile video, but it has been difficult to customize physical layer functionalities by existing SDKs on PC. Therefore, in this paper, we utilize GNU Radio, which has been widely used as a development tool of the software radio, and evaluate the performance of VLC streaming while changing modulation, transmission gain, and etc.

We have introduced a proactive content delivery system utilizing transportation systems for next generation mobile networks. In this paper, we propose a proactive delivery scheduling, which is called "Smart Scheduler." This scheduler calculates contents quality and quantity and also selects delivery location and timing. To validate and evaluate our method, we experiment in the real environments. Experimental results conduct that our method performs effectively.

We have already introduced an analytical model for Comfort Route (CR) Navigation providing high communication quality routes to users. In this paper, to estimate CR performance in more realistic environment, we modify the previous analytical model to Heterogeneous Network (HetNet). To validate our model, we employ several scenarios in real experiments. We evaluate throughputs characteristics at ten public Wi-Fi spots in Takadanobaba, which is a city near to our university. We compare the results in real experiments and prove that our model works well with high accuracy.

In video coding, image quality degradation is inevitable due to quantization steps in the encoding process. We had proposed a blind type (no-reference type) PSNR estimation method which uses SVM and natural image statistics. In this paper, as an application of the blind PSNR estimation, we propose a method to predict PSNRs of video sequences before applying video compression by using image features of the original video sequences. As a result, we reveal that we can predict sufficient PSNRs for more than 80% frames within 5dB prediction errors.

International standardization of multi-view and 3D video coding formats is now the hot topic at MPEG. For the latter, as an encoder input, video signals and depth maps are utilized simultaneously for efficient compression. Moreover, recently, video sequence with a depth map can be easily taken by an RGB-D (depth) sensor like Kinect. In this paper, we try to improve picture quality of H.265/HEVC compressed streams by using depth map. Concretely, we separate video signals into foreground and background, and detect the position of foreground regions by using this information. Furthermore, we detect boundary regions of video signals by using the depth map, and we achieve quality improvement of H.265/HEVC.

Generally, PSNR (Peak Signal-to-Noise Ratio) is used as an index of the image quality objective evaluation, and it is necessary to prepare an original image. However, it is usually difficult to prepare the original images in case of CGM (Consumer Generated Media) movies such as YouTube etc. The authors have been working on blind PSNR estimation which estimates PSNRs of compressed and decoded videos without using original images and bit-stream analysis. In our previous study, we have estimated the quality of all frames to investigate its estimation accuracy, but we focus on quasi real-time implementation of our approach in this paper. In contrast with our conventional method, we sample the results of our blind PSNR estimation at a certain interval and interpolate PSNRs of other frames based on their AC power levels.

In this paper, we describe a method to accelerate image mosaicking for omnidirectional in-vehicle camera images. In this research, it is assumed that omnidirectional in-vehicle camera images are taken from six cameras (front, back and each sides) on the car. We want to mosaic images to create an all-around images and, in order to accelerate the procedure, we extract two steps which are independent of each other. One is estimation of a camera matrix from images, and the other is image composition by the camera matrix. By using the camera matrix that was estimated for a previous frame to a current frame, we skip the camera matrix estimation process. Furthermore, by applying a correctly estimated camera matrix for an image pair to a failed pair, we confirmed improvement of the compositing results.

Improving Quality of Service (QoS) in wireless networks is important and necessary for mobile users. We have previously proposed Comfort Route (CR) Navigation, which navigates users to their destinations using high QoS communication areas, such as Wi-Fi APs, rather than the geographical Shortest Route (SR). In this paper, we employ an analytical model to estimate the CR gain in a theoretical manner which assumes that available cellular and Wi-Fi throughputs are uniform within their coverage. The CR gain is computed by using basic parameters, including wireless network bandwidth and transmission time. To validate our model, we compare simulation results and real observation. These results conclude that the CR gain could estimate by using our analytical model.

To improve Quality of Service in wireless networks while mobile users travel, we introduce Comfort Route Navigation (CRN) which is a navigation system based on a user centric mobility management for next generation wireless networks. CRN provides an optimal route that satisfies user needs, such as obtaining maximum wireless resources. To achieve CRN, we construct an access point (AP) map of Shinjuku city. To reflect quality of APs in this map, we evaluate throughputs at seven public Wi-Fi spots. Based on these observations, we evaluate performance of our CRN via computer simulations and in a real environment. These evaluations conduct that a CRN user could obtain higher communication quality rather than a Shortest Route (SR) user. These evaluations also conclude that CRN gain depends heavily on the quality and location of the best broadband spot.

In recent years, video sequence with a depth map can be taken by a camera with RGB-D (depth) sensor like Kinect. On the other hand, quality degradation at occluded regions is inevitable when conventional 2-D video coding is applied. In this paper, we propose usage of an occluded region detection supported by the depth map for improvement of 2-D image coding quality. Experiments show that further improvement of picture quality at the whole of frame and occluded regions is addressed for H.265/HEVC compressed streams. © 2014 IEEE.

It is important for multimedia mobile users to obtain sufficient resources and a stable capacity of resources. Network resources such as available bandwidth for mobile users, however, fluctuate depending on the level of background traffic. The deployment of several access points enables mobile users to access more network resources on the way to their destination by taking the longcut route rather than the shortcut route. These users pay an extra cost to take the longcut route but obtain greater gains in throughput over the shortcut route. Previous research has shown that the gain-to-cost ratio can be more than double for the longcut route. No considerations, however, have been made for the case in which resource fluctuation is due to network congestion. This paper investigates the quality of service (QoS) for a longcut route with fewer resources (real resources) compared with that for one with the total resources (ideal resources) in the longcut route calculation. The results show that on average, users have gains of approximately double in real resources than their gains in ideal resources. This finding indicates that users can greatly improve their gains if they can accurately predict the resource fluctuation.

The peak signal-to-noise ratio (PSNR) used as an index of image quality usually requires original images, but this is difficult for consumer generated content such as videos on YouTube. Therefore, we developed two blind PSNR estimation methods without bit-stream analysis in which multiple support vector machines are prepared to learn differently encoded images in PSNR
using an entire frame and dividing the frame into two areas. We confirmed that higher estimation accuracy is possible for the latter method against that using the entire frame.

With the increasing demand for video streaming in mobile environment, we have been tackling a research on throughput prediction based on the measurement history data in order to improve service quality[1][2]. Since throughput fluctuation occurs sharply in mobile environment, useless packet loss or congestion occurs when available network bandwidth is insufficient compared to the bit rate of video contents. Hence, it is possible to improve the quality of service by deducing the available throughput and selecting appropriate bit rate for the video contents. In this paper, we propose a throughput prediction method that considers mobile locations of a train and their throughput history and show promising experimental results.

Extending battery life for smartphones while using wireless networks is important. In this paper, we propose Comfort Route for Energy Saving (CRFES), which helps users extend the battery life of their smartphones by navigating to their destinations via Quality of Service (QoS)-compliant energy-efficient spots such as Wi-Fi spots. To create CRFES, we construct maps of Wi-Fi spots using logging software we developed for Android phones. In addition to locations, we record actual throughput and energy consumption observed in cellular/Wi-Fi networks. So comfort route navigation is done by choosing a route which minimizes energy consumption instead of choosing the shortest path. Evaluations are carried out by computer simulation and one real city map. The results conclude that CRFES can save the battery life by approximately 50% by spending 1.6 times time cost to reach their destinations on average.

Traffic sign recognition contributes to the safety of drivers and people around the car. The system analyzes the road ahead images taken by on-vehicle camera, but blurred and distorted images make it difficult to recognize correct traffic sign. In this paper, we propose a method that deforms from ellipses to perfect circles and applies template matching with distorted sign templates.

Height data is very important not only as an indicator of growth in low birth weight infants, newborn, infants (infants below), but also for early detection of disease such as dwarfism. In this paper, we consider height estimation by image processing that targets lying infants for the purpose of reducing the burden of infants, doctors and parents. Furthermore, we compare results of using a single image and multiple images obtained by moving pictures, and show the latter contributes to improvement.

Feature analysis of audio compression is necessary to achieve high accuracy in musical content recognition and content-based music information retrieval (MIR). Bit rate differences are expected to adversely affect musical content analysis and content-based MIR results because the frequency response might be changed by the encoding. In this paper, we specifically examine its effect on the chroma vector, which is a commonly used feature vector for music signal processing. We analyze sound qualities extracted from encoded music files with different bit rates and compare them with the chroma features of original songs obtained using datasets for chord recognition. © 2014 Springer International Publishing.

Providing robust content delivery along with efficient wireless resource usage is important for next generation wireless networks. To achieve this, we propose a proactive content caching scheme utilizing transportation systems, especially trains. In our system, we place content servers with CCN capability to every train and station. Segments of video contents are pre-cached by the station servers before trains arrive at stations. Trains receive the contents via high-speed wireless transport while they stop at the stations. We develop a prototype system based on IP and CCN Hybrid protocols. We evaluate its performance by field experiment and compare with traditional CDN scenarios using cellular networks. Evaluations conclude that our system can achieve high-speed and high-reliable video delivery without freezing.

Recently, extending battery life of mobile devices is important for mobile users. We have already introduced Comfort Route (CR) Navigation with energy saving. In this paper, to estimate energy saving effect in a theoretical manner, we employ the analytical model for this route. We find that the effect depends on four basic parameters, such as cellular and Wi-Fi throughput, energy consumption for cellular, Wi-Fi, and idle states, CR and Shortest route cellular usage ratio, and CR longcut ratio. To validate our model, we compare to simulated results and prove that our model works well with high accuracy.

With the increasing demand for the energy-efficient communications, we have been tackling the research on the efficient video delivery and caching technologies over ICN (Information Centric Network) as one of the research items of the GreenlCN (Architecture and Applications of Green Information Centric Networking) project. One way to make the video delivery more efficient would be to cut down on the packet losses during the delivery. Since the actual available throughput that is lower than the video delivery rate causes the packet losses, the efficient video delivery can be realized by deducing the actual throughput. In this paper, we propose the mobile performance prediction method where the throughput is predicted based on the two criteria. One is the throughput that is estimated based on the observed radio signal quality. The other is the estimated throughput based on the history of the connected base station and time to access. In addition, we show some results of the two months-long experiment measuring hourly fluctuation in throughput.

Recently, improving Quality of Service (QoS) in wireless networks is important for mobile users. We have already introduced Comfort Route (CR) Navigation providing high communication quality. In this paper, to estimate CR performance in a theoretical manner, we employ the analytical model for CR. To validate our model, we compare to simulated results and prove that our model works well with high accuracy. We find that the CR gain depends on four basic parameters, such as the cellular and Wi-Fi throughput ratio, the shortest route Wi-Fi coverage area, the CR cellular usage ratio, and user cost which specifies longcut margin. We also find that the best CR gain strongly depends on the cellular and Wi-Fi throughput ratio.

The threat of malware is increasing. Malware detection has thus become important for the safety of the Internet usage. In this paper, our goal is to classify Internet traffic into normal or anomalous class correctly to prevent malware from spreading through the Internet. To stop malware, we are focusing on the time series changes between normal and malware traffic. We propose a malware detection method by using N-gram to handle time series changes. Classification of Internet traffic into normal or anomoalous class is done by using posterior probability calculated by nearest neighbor density estimation. We use CCCDATAset2010,2011 as malware traffic, and application traffic captured dby a PC as normal traffic.

Video streaming is one of the most popular Internet applications for years. To improve content availability and scalability, CDN and P2P networking have been widely used. However, the current Internet architecture, which relies on a host-based communication, has no ability to leverage caching at the network level. Recently, Content Centric Networking (CCN)has been emerged as a new networking paradigm. The communication of CCN focuses on the named contents instead of the IP addresses. This paper provides a framework for enabling video streaming over CCN. We performed experiments using CCNx on the PlanetLab to explore how an overlay path and a chunk size affect the streaming quality. Furthermore, we study the deployment of MPEG-DASH to achieve adaptive bit-rate streaming over CCN, which reveals additional challenge.

Previously, multimedia streaming was mainly operated by RTP/UDP and RTSP, and HTTP streaming was used only to pass through the firewall. However, with rapid growth of broadband network and CDN, HTTP streaming becomes much popular recently as represented by YouTube. Furthermore, HTTP Live Streaming by Apple had attracted a lot of attention, and then MPEG-DASH has been specified as an international standard for the HTTP streaming. On the other hand, LTE (4G) networks emerge and deploy in a rapid pace all over the world. Therefore, in this paper, we evaluate performances of HTTP Live Streaming and MPEG-DASH under handover events in LTE networks.

Underwater Sensor Networks have received growing interests recently, which can monitor and collect underwater environments. However, it is difficult to transmit data in underwater environments because of some effects such as high propagation delay. noises. Underwater Sensor Networks need routing protocols similar to radio propagation in terrestrial. However, existing routing protocols are not effective in Underwater Sensor Networks because of various influences listed above. We focus on routing protocols to improve network performance.DTN (delay/disruption tolerant network)assumes high propagation delay and intermittent link. In this paper, Epidemic Routing [2], which was proposed for DTN, is used as a routing protocol for Underwater Sensor Networks and its effectiveness against AODV is investigated.

Underwater sensor networks constructed in ocean are attracting many researchers' interests as an effective method to collect underwater environmental information. In these networks, Autonomous Underwater Vehicles (AUV)equipped with sensor devices and acoustic modems are used, and many attentions have to be paid to decrease their energy consumptions. In this paper, we make formulas of energy consumptions for next two scenarios, sending data by acoustic waves and moving AUVs, by referring to related works. Moreover, we carry out some simulation experiments by using network simulator NS-3, and clarify methods to decrease energy consumptions in an efficient manner.

Along with evolution of the broadband network, various services appear on the Internet. However quality of services (QoS) often gets worse because increasing users place heavy load on the network and servers. To solve this problem, it is necessary to build the high performance and stable network. But, it is difficult to analyze jitters because there is no approximation formula for the variance of waiting time in queuing systems of the network. In this paper, we consider the approximation formula in a FIFO(First-In-First-Out)GI/GI/1 queuing system and evaluate its approximation in accuracy in the actual environment. We do comparison verification of our approximation formula by a simulation and measured traffic.

Recently, the demand for visualizing congestion levels of Wi-Fi spots is increasing because Wi-Fi spots are said to be quite congested. In this paper. for this purpose, we employ a simple approximation formula about the relationship between throughputs and the number of active users associated with the access point. In addition, we try to estimate fiiture throughput variation which depends on fluctuation of the associated users using a state transition model. Then, we classify the congestion level into four categories depending on the current and future expected throughputs and evaluate the accuracy of our proposal. We found that our proposal shows acceptable accuracy rate of 70〜90% in our experimental conditions.

PSNR needs original pictures, but this is difficult for CGM contents like YouTube. Therefore, we had developed a blind PSNR estimation method in which multiple SVMs are prepared to learn differently encoded images in PSNR. We also tried a method which divides a frame into two regions by using. Saliency Map to estimate PSNR per region. However, this approach sometimes fails because it always tries to extract salient regions in a relative manner even if images are totally flat. Therefore, in this paper, we consider a method which uses AC levels of local small regions to separate input images instead of Saliency Map. Then, we confirm our method provides more stable results even when the previous method fails.

In video coding using block transforms, blocking artifacts often occur in compressed images especially at low bit rates. Although PSNR and SSIM are often used as objective quality measures of compressed images, they cannot fully evaluate degradation of image quality affected by the blocking artifacts. In this paper, based on the results of a number of subjective evaluations collected for MPEG-2 compressed image sequences, we evaluate the correlation between image features and subjective evaluations of blocking artifacts. We also investigate an NR-type evaluation method of the blocking artifacts as an application of the above results.

Image super-resolution quality degrades by noise component. However it should not be eliminated because important for high definition impression for digital cinema. Therefore we propose an image super-resolution method by synthesis of super-resolution signal and noise components respectively after dividing an original image into signal and noise components. The signal component is resolution-enhanced using registration between the signal component and its wavelet multi-scale components with resolution-enhanced parameter optimization.

In this paper, we show characteristics of geomagnetic data and acceleration data captured by a three-axis geomagnetic sensor and a three-axis acceleration sensor in a mobile terminal, and suggest effectiveness of sensor data of the mobile terminal to estimate person mobility. Person mobility means how a person moves from one point to another, and we check four types of person mobility, walking, by train, by bus and by car. As a result of investigation, magnetism in the train and accelerations for each driving situation in the bus or in the car will be characterized. We also investigate effectiveness of statistical features, average and standard deviation, and power spectrum for human behavior classification like walk, run, stop, sit and stand.

In this paper, we show characteristics of geomagnetic data and acceleration data captured by a three-axis geomagnetic sensor and a three-axis acceleration sensor in a mobile terminal, and suggest effectiveness of sensor data of the mobile terminal to estimate person mobility. Person mobility means how a person moves from one point to another, and we check four types of person mobility, walking, by train, by bus and by car. As a result of investigation, magnetism in the train and accelerations for each driving situation in the bus or in the car will be characterized. We also investigate effectiveness of statistical features, average and standard deviation, and power spectrum for human behavior classification like walk, run, stop, sit and stand.

Three-dimensional (3D) video service is expected to be introduced as a next-generation television service. Stereoscopic video is composed of two 2D video signals for the left and right views, and these 2D video signals are encoded. Video quality between the left and right views is not always consistent because, for example, each view is encoded at a different bit rate. As a result, the video quality difference between the left and right views degrades the quality of stereoscopic video. However, these characteristics have not been thoroughly studied or modeled. Therefore, it is necessary to better understand how the video quality difference affects stereoscopic video quality and to model the video quality characteristics. To do that, we conducted subjective quality assessments to derive subjective video quality characteristics. The characteristics showed that 3D video quality was affected by the difference in video quality between the left and right views, and that when the difference was small, 3D video quality correlated with the highest 2D video quality of the two views. We modeled these characteristics as a subjective quality metric using a training data set. Finally, we verified the performance of our proposed model by applying it to unknown data sets.

In Peer-to-Peer (P2P) networks, an incentive mechanism is a necessary component to deal with the free-riding behavior. The challenge is that direct reciprocal incentives
e.g., tit-for-tat, which consider the cooperation of peers in a pair-wise manner, are not suited with P2P streaming. In this paper, we propose a new service differentiation mechanism to provide a redistribution incentive for P2P streaming in a hybrid overlay network. The contribution of a peer can be measured from the number of video sub-streams that it uploads to other peers. By sending one request message, the number of sub-streams that each peer can retrieve is varied by its contribution level. An altruistic peer thus has to send less request messages and will experience smoother video quality than a selfish peer. Through simulations, we demonstrate that our solution can provide service differentiation among peers with better streaming quality than the tit-for-tat scheme.

Network resources for mobile users such as available bandwidth fluctuate depending on the level of background traffic. The deployment of many access points enables mobile users to access more network resources on their way to their destination by taking the longcut route rather than the shortcut route. These users pay some extra cost to take the longcut but obtain greater gains, such as in throughput, over the shortcut route. Previous research has shown that the gain-to-cost ratio can be more than double for the longcut route. No considerations, however, have been made for the case in which resource fluctuation is due to network congestion. This paper investigates the Quality of Service (QoS) for a longcut route with fewer resources (real resources) compared with that for one with the total resources (ideal resources) in the longcut route calculation. The results show that, on average, users have gains of approximately twice in real resources of their gains in ideal resources. This finding indicates that users can greatly improve their gains if they could accurately predict the resource fluctuation. Copyright © 2013 ACM.

This paper presents an RTT-fair TCP congestion control using ACK interval measurement and extends the approach to have inter-protocol friendliness, especially with CUBIC TCP. In the previous RTT-fair TCP congestion control, including ours, estimation of RTTs of a competing flow had been a problem. We try to solve this problem by measuring ACK arrival intervals, which are observable parameters by an end host. This approach enables estimation of congestion window behaviors, in addition to RTTs, of a competing flow. We then extend our congestion control to have friendliness to CUBIC-TCP, in addition to classical TCP-Reno, in an adaptive manner. Extensive experiment results are shown for simulations and implementations and effectiveness of our approach is confirmed.

This paper presents a TCP differentiation based on TCP version identification and IEEE 802.11e EDCA for low delay multimedia streaming over wireless LAN. It has been known that delay-based TCP can achieve low delay transport as long as it does not compete with loss-based TCP. However, when competition happens, it seriously decreases its rate by itself due to RTT increase. In order to alleviate this problem, we consider combination of TCP version identification and prioritized transport of delay-based TCP by EDCA at an access point. We evaluate this approach by simulations and implementations, and confirm its effectiveness. © 2013 IEEE.

In vehicle-to-vehicle communication, a message named BSM (Basic Safety Message) has a major role to inform a driver about surrounding condition. A vehicle periodically sends the BSM which includes the information of itself. Traffic load issues caused by BSMs would often be raised on heavily congested roads. Therefore, BSM congestion controls to avoid traffic congestion are challenging. This paper proposes a new BSM congestion control named BRAEVE. BRAEVE adapts its BSM generation rate according to the number of neighbor vehicles in communication range and then controls overall network traffic load. Our simulation evaluation shows that BRAEVE provides more uniform recognition of surrounding vehicles than an existing method, which brings much better safety assurance to a driver.

Objective picture quality measures cannot estimate the effect of blocking artifacts caused by video compression sufficiently. In this paper, we apply cepstrum analysis to quantify the blocking artifacts. We show experimental results for some test sequences using different coding schemes and prove effectiveness of our approach.

Quality of image super-resolution (SR) degrades by noise component. However it should not be eliminated because it is important for high definition impression for digital cinema. Therefore we propose an image SR method by synthesis of resolution-enhanced signal and noise components respectively after dividing an original image into signal and noise components. The signal component is resolution-enhanced using image registration between the signal component and its wavelet multi-scale components with resolution-enhanced parameter optimization. © 2013 IEEE.

Most network applications nowadays are related to the distribution of contents from a source to a number of users. Several developments such as CDN and P2P networking have been used to provide access to contents at the network edges. However, they have no ability to leverage caching at the network level and still rely on a host-based communication model. Recently, Content Centric Networking (CCN) has been emerged as a new networking paradigm. In CCN, the communication focuses on the named content instead of the IP address. This paper provides a framework for enabling video streaming over CCN. We performed experiments using CCNx on the PlanetLab to explore the effects of an overlay delivery path and a chunk size to the streaming quality. Furthermore, we study the deployment of MPEG-DASH to achieve adaptive bit-rate streaming over CCN, which reveals additional challenge. © 2013 IEEE.

We present a robust cellular network (RoCNet) that combines a cellular and an opportunistic networks for spatial uplink mobile data offloading, which focuses on the spatial difference of the traffic load among areas (e. g., business district and residential area in the daytime). RoCNet realizes the spatial data offload by leveraging the store-carry-forward routing mechanism. In the area where traffic load is high, delay-tolerant data originated from a mobile terminal is directly forwarded to a nearby terminal using Bluetooth or wireless LAN instead of being transmitted to a congested cellular base station. When the data is carried by the nearby terminal to other area where the traffic load is low, the data is forwarded to a cellular base station. To enhance the offload effect, it is necessary for data to be forwarded to a terminal that moves to a low traffic load area. In this paper, we use the particle filter to predict user behavior. Before forwarding data between mobile terminals, the terminals exchange prediction results and decide whether or not the data should be forwarded. We conducted a computer simulation whose result shows RoCNet can spatially offload uplink traffic in a traffic concentration area to non-congested areas. As a result, RoCNet can suppress peak traffic by about 20 percent in a traffic-congested base station by distributing traffic to vicinity base stations.

Recently, Internet services and wireless infrastructures have grown widely and diversely, and people can easily access Internet anytime and anywhere. Smartphones have become an indispensable part for this purpose, so people are very nervous about smartphone's battery life. In this paper, we therefore propose Comfort Route for Energy Saving (CRFES) which helps users reach their destinations while stopping by Quality of Service (QoS) fulfilled low-power communication places such as Wi-Fi spots. We evaluate the performance of CRFES in a computer simulation and a real environment. We find that CRFES can save mobile device batteries drastically by using Wi-Fi rather than cellulars with small extra time cost to visit hot spots.

Network traffic volumes continue to increase rapidly with spread of the broadband access and increase of Internet backbone capacities. Moreover, services provided over the Internet become more diverse. In order to prevent network traffic congestion, it is necessary to estimate network traffic volumes so that network administrators can prevent deterioration of quality of services by estimating and regulating traffic in advance. Therefore, we study a method to estimate network traffic volumes with external factors, for example, calender, weather data, and so on. In this paper, we suggest a method for estimating network traffic using logistic regression, and prove its effectiveness by comparing existing method with ours.

The demand for access to communication while in transit is increasing, especially among mobile device users. Previous our research has introduced an optimal route problem and a number of solutions to this problem. However, to date, the literature has only presented evaluations based on a theoretical model. In this paper, to validate and enhance the previous work, we observe actual throughput characteristics in public wireless LAN services such as FreeSpot and DOCOMO 's Mzone. We evaluate an optimal longcut route navigation using the results of these observations and show its effectiveness in computer simulations and experiments in actual environment. In both experiments, the optimal longcut route can gain data amount approximately 15 times greater than that of a short cut route.

Conventional TCP protocols were initially designed for wired networks and increase their window sizes aggressively. Therefore, they cause heavy contentions and packet collisions in MAC layer over wireless multi-hop ad-hoc networks. A congestion control method extending TCP Vegas was proposed for such networks. This method has low-delay property and works effectively in simple environments. However, this method does not work effectively in realistic and complicated environments. In order to improve its throughput performance, we adjust its packet loss handling and prove its effectiveness by simulations along with upper bound of TCP throughput over wireless multi-hop ad-hoc networks.

This paper presents a TCP differentiation combining TCP version identification and EDCA control . In current proposed TCP congestion controls, TCP Vegas controls congestion window (cwnd) based on round trip time (RTT), and it can suppress RTT. Therefore, TCP Vegas can do efficient communication alone. However, TCP Vegas decreases its throughput when it competes with other TCP versions. To solve this problem, in our previous paper, we proposed the TCP differentiation and it brings much better throughput in simulation experiments. In this paper, we present the advantages of this TCP differentiation in both implementation and simulation experiments assuming real environment.

Underwater sensor networks constructed in the ocean have attracted many researchers' interests these days. In these networks, Autonomous Underwater Vehicles (AUV) with sensor devices and modems equipped are used to collect underwater information. In such an environment, to reduce energy consumption is very important. In this paper, two different types of energy consumption, data transmission and AUV movement, are compared by calculating formulas and carrying out simulations.

Recently, many papers focus on Underwater Acoustic Sensor Networks (UWASNs) to collect the data in underwater environments. However, the data transmission in these underwater environments suffers from various influences such as bandwidth usage limitation, surrounding noise, and large acoustic propagation delays. Therefore, the communication over UWASNs is more difficult than radio propagation in terrestrial. In many researches, protocols at the link layer and the network layer for UWASNs have been proposed for decades. We focus on an automatic repeat-request (ARQ) and IEEE 802.11 CSMA/CA to reduce contention for transmission over UWASNs. In this paper, we propose a MAC protocol with Continuous and Selective ARQ and define a new rule of NAV/CW for UWASNs, which is called Synchronous NAV and Linear CW. Through ns-2 based simulator, we show effects and differences of the network performances by using our MAC protocol.

Recently, ITS (Intelligent Transportation System) technology has developed significantly. In vehicle-to-vehicle communication of ITS, a message named BSM (Basic Safety Message) has a major role to inform a driver about dangerous situation before accidents happen. A vehicle periodically sends the BSM which includes the information of itself. Traffic loads caused by BSMs are raised on heavily congested roads, for example on a highway. Therefore, in order to avoid traffic congestion by BSMs, methods to change generation rates or transmission powers of BSMs are discussed in recent study. This paper focuses on the rate control of BSM and points out shortcoming of conventional methods in more realistic environment. We propose a solution to the issue and show simulation experiments.

When we try to improve fairness between TCP connections, estimating network conditions is effective. For example, we can collect network information through relay hosts. But TCP congestion control mechanism is autonomously distributed control in current networks like Internet. So we should use information which terminals can measure. In this paper, we estimate network conditions by measuring ACK intervals and propose congestion control methods which have RTT fairness and friendliness to new standard TCP by using the information.

We have started designing a robust cellular network (RoCNet) that combines infrastructure-based (e.g., cellular) and opportunistic networking for disaster controlling and spatial data offloading. The RoCNet provides communication means without an infrastructure network in some cases such as at the time of disaster by using a store-carry-forward fashion as in the Delay Tolerant Network (DTN). When the infrastructure network is available, the RoCNet facilitates the traffic offloading leveraging the store-carry-forward feature considering a degree of traffic congestion, traffic direction, etc. In this paper, we give an overview of the RoCNet concept and show a simulation result as a proof-of-concept work. The result shows RoCNet can spatially offload uplink traffic in the traffic concentration district to the non-congested area.

We will apply cepstrum analysis, which is often used as a voice recognition method, to classify walking states (gait) using three dimensional acceleration sensor and examine this validity. Then we will show that cepstrum gives us better recognition and is less susceptible data than basic statistics such as average and variance against the data acquisition condition. Also we will study characteristics of spectral envelope of walking states.

This paper focuses on RTT-fairness of multiple TCP flows over the Internet, and proposes a new TCP congestion control named "HRF (Hybrid RTT-Fair)-TCP". Today, it is a serious problem that the flows having smaller RTT utilize more bandwidth than others when multiple flows having different RTT values compete in the same network. This means that a user with longer RTT may not be able to obtain sufficient bandwidth by the current methods. This RTT fairness issue has been discussed in many TCP papers. An example is CR (Constant Rate) algorithm, which achieves RTT-fairness by multiplying the square of RTT value in its window increment phase against TCP-Reno. However, the method halves its windows size same as TCP-Reno when a packet loss is detected. This makes worse its efficiency in certain network cases. On the other hand, recent proposed TCP versions essentially require throughput efficiency and TCP-friendliness with TCP-Reno. Therefore, we try to keep these advantages in our TCP design in addition to RTT-fairness. In this paper, we make intuitive analytical models in which we separate resource utilization processes into two cases: utilization of bottleneck link capacity and that of buffer space at the bottleneck link router. These models take into account three characteristic algorithms (Reno, Constant Rate, Constant Increase) in window increment phase where a sender receives an acknowledgement successfully. Their validity is proved by both simulations and implementations. From these analyses, we propose HRF-TCP which switches two modes according to observed RTT values and achieves RTT fairness. Experiments are carried out to validate the proposed method. Finally, HRF-TCP outperforms conventional methods in RTT-fairness, efficiency and friendliness with TCP-Reno.

Previous our research has introduced an optimal longcut route problem and a number of solutions to this problem. However, to date, the literature has only presented evaluations based on a single theoretical radio propagation model. Because throughput characteristics derived from radio propagation models significantly affects longcut route performance, this study introduces three simple models by referring to actual measurement values. The results of our evaluation show that there are large differences in performance between the three models. The Non-Line-of-Sight model can have an improvement ratio approximately five times greater than that of the Line-of-Sight Indoor model and six times greater than that of the previous theoretical model. © 2012 IEEE.

In Peer-to-Peer (P2P) networks, a great number of peers are reluctant to contribute to the system. An incentive mechanism thus becomes a necessary component to motivate the cooperation of peers. Moreover, due to the continuous nature of data transmission, the P2P media streaming requires different incentive strategies from the P2P file-sharing. In this paper, we propose a new service differentiation mechanism to provide a redistribution incentive for P2P streaming in a hybrid overlay network. The contribution amount of a peer can be measured from the number of video sub-streams it transmits, instead of the quantity of transferred data. By sending one request message, the number of sub-streams that each peer can retrieve is varied by its contribution amount. An altruistic peer thus has to send less request messages and experiences smoother video quality than a selfish peer. Through simulations, we demonstrate that the proposed mechanism can provide service differentiation among peers according to their contribution amount. © 2012 IEEE.

The growing popularity of mobile data communication has been leading a growing lack of radio resource on a cellular network. Therefore, to avoid the data traffic from exceeding the capacity limit of a cellular base station (BS), cellular operators have been trying to install cellular femto cells (local BSs) or wireless LAN access points (APs) in restaurants, cafes, user's houses, etc., to divert data traffic from the cellular radio access link to the fixedline such as x-DSL and FTTH. However, the traffic levels extremely change in a day, even locally, due to the mobility of users. While peak data traffic at certain times of a day appears locally, cellular BSs as well as the femto cells and the APs would sit underutilized for the rest of the day. Thus, we have started designing a robust cellular network (RoCNet) that combines infrastructure-based network (e.g., cellular network) and opportunistic networking for spatial mobile data offloading, which focuses on the difference of the data traffic level among areas (e.g., business district and residential area). The RoCNet facilitates the data traffic offloading leveraging the storecarry-forward feature considering a degree of traffic congestion. In this paper, we give an overview of the RoCNet and show a simulation result. The result shows RoCNet can spatially offload uplink traffic in a traffic concentration area to non-congested areas. In other words, RoCNet can suppress peak traffic in a traffic-congested base station by distributing traffic to vicinity base stations. Copyright © 2012 ACM.

Bit-depth scalable coding method is an approach that generates multiplexed bit-streams that can be decoded to two video sequences, for Standard Dynamic Range (SDR) environment and for High Dynamic Range (HDR) environment. This paper presents a bit-depth scalable coding method that is considered as gradation restoration in inter-layer prediction process. Our proposed inter-layer prediction uses a histogram interpolation method that enables to generate more mid-gray brightness levels than traditional inverse tone mapping methods with one-to-one correspondence.

This paper presents a chord recognition method from music signals using chroma vectors and musical knowledge known as Doubly Nested Circle of Fifths (DNCOF). DNCOF represents the relationships of major and minor chords where the neighboring two triads are similar. We obtain a novel feature from chroma vectors by mapping them onto two-dimensional DNCOF coordinate, which we call DNCOF vectors. We expect that the DNCOF vectors can contribute to correcting false recognition obtained by the chroma vectors when their mapped positions are apart from one another in the DNCOF coordinate. In this research, we evaluated our proposal using the Beatles' datasets and showed its effectiveness.

Feature extraction of images can be applied to image matching, image searching, object recognition, image tracking etc. One of the effective methods to extract features of images is Scale-Invariant Feature Transform (SIFT) [1]. In this paper, we indicate problems of SIFT and propose a method to improve its performance by applying Bilateral Filter [2]. In addition, we implement is acceleration by GPGPU (general purpose GPU), apply this method to generic object recognition and perform a comparison experiment. We compare the proposed method with the original method using SIFT and confirm improvement of the identification rate by the proposed method.

We propose a novel wavelet domain image super-resolution method from digital cinema to ultrahigh definition television considering cinema noise component. The proposed method features that spatial resolution of an original image is expanded by synthesis of super-resolved signal and noise components respectively after dividing an original image into signal and noise components. Dividing noise component uses spatio-temporal wavelet decomposition based on frequency spectrum analysis of cinema noise. And super-resolution parameters are optimized by comparing size-reduced super-resolution images with an original image. Experimental results showed that a super-resolution image using the proposed method has a subjectively better appearance and an objectively better peak signal-to-noise ratio measurement than conventional methods.

This paper presents a music part detection method incorporating chroma vector analysis for use with music TV programs. Results show that envelopes of chroma components of music signals tend to have horizontal (i.e. temporal) correlation in time-frequency representation because music signals have a periodic chord sequences. Based on this fact, we analyze time series of chroma components and attempt to segment music parts in music TV programs from other parts. Experimental results show an F-measure of 0.78, which is better than that obtained using the previous method.

We have proposed to take a "longcut" route by which users can communicate more effectively than a "shortcut" route when they move from a source to a destination. In our previous study, by taking permissible time consumption into account as a parameter, we evaluated the effectiveness of an optimal longcut route in the circumstance where users ideally utilize the wireless LAN resources, but network failures and congestions usually occur in real networks. In addition to the previous study, this paper is devoted to evaluate the effectiveness of an optimal longcut route considering bottleneck existence by network congestion. As the result, considering bottleneck information, we present that the optimal longcut route can get about twice effectiveness against the previous study.

Recently, fast and broadband networks are deployed all over the world and, in parallel, new communication methods which have not been sufficiently investigated appears such as underwater sensor networks. Live video streaming services had been conveyed over RTP/UDP stacks but, recently, they are transmitted over TCP stacks such as YouTube and Ustream. This paper introduces research topics about the transport protocols which have been designed for wired networks, wireless networks and underwater sensor networks. Furthermore, this paper also refers to the direction toward integrated design of transport protocols in future.

Previously, multimedia streaming was mainly operated by RTP/UDP and RTSP, and HTTP streaming was used only to pass through the firewall. However, with rapid growth of broadband network, HTTP streaming becomes much popular recently as represented by YouTube. Furthermore, HTTP Live Streaming proposed by Apple has attracted a lot of attention along with the popularization of iPhone and iPad. Therefore, in this paper, we evaluated TCP performances for HTTP Live Streaming. Since various TCP versions were proposed so far, we compare TCP-Reno as a loss-based protocol, TCP-Vegas as a delay-based protocol, CUBIC-TCP as Linux default, which is another loss-based protocol, and TCP-Fusion as a hybrid TCP for evaluation.

Recently, performance of the Internet has been improved, and the video coding technologies have also evolved. As a result, many various streaming services have been provided. Especially, many people have had a chance to enjoy these services using their mobile phones such as iPhone. Therefore, we need to focus on specific researches on video streaming over wireless networks. In this paper, we implement VTP (Video Transport Protocol) and Hybrid-TFRC as adaptive rate control mechanisms in VIC (Video Conferencing Tool). And we evaluate their performances using Linux notebooks in various environments. We observe the effectiveness of these implementations and the fairness evaluation when competing with a TCP-Reno flow over wired and wireless combined networks.

The conventional TCP protocol was initially designed for wired networks and increase aggressively its window size. Therefore, the contention and packet collision in MAC layer are occurred over wireless multi-hop ad-hoc networks. A congestion control method extending TCP Vegas was proposed for such networks. However, this method does not work effectively in some environments. In order to improve its throughput performance, we apply the Adaptive Packet Size method that changes its packet size adaptively.

This paper presents a method of TCP version identification for TCP version differentiation. Among current proposed TCP congestion controls, TCP Vegas controls its congestion window (cwnd) based on round trip time (RTT), and it can suppress unnecessary RTT increase. Therefore, TCP Vegas can provide efficient transport by itself. However, TCP Vegas decreases its throughput when it competes with other TCP versions. To solve this problem, we identify TCP versions at a router or a Network Interface Card (NIC), and preferentially transmit TCP Vegas flows and differentiate TCP versions.

Recently, wireless LAN based on IEEE 802.11(WiFi) is widely used due to its easy deployment and high-speed connection. As the devices become smaller, higher mobility is required. Thus, it is assumed that handovers, which occur when the device try to change its access point, happen frequently. Furthermore, realtime applications such as VoIP and video conference are popularly used. However, in general, it is said that handovers over WiFi environment take a few hundreds of milliseconds. During handover process, 90% of overall handover delay is caused by channel scanning phase. Therefore, it is essential to reduce channel scanning latency and, in this paper, a new approach to reduce channel scanning time is proposed and analyzed.

Recently, many papers focus on the communication over Underwater Acoustic Sensor Networks (UWASNs) to collect the data in underwater environments. However, the data transmission in these underwater environments suffers from various influences such as bandwidth usage limitation, surrounding noise, and large acoustic propagation delays. Therefore, the communication over UWASNs is more difficult than radio propagation in terrestrial. In many researches, protocols at the link layer and the network layer for UWASNs have been proposed for decades. We focus on an automatic repeat-request (ARQ) to ensure reliability of the communication. The simplest ARQ protocol is the Stop-and-Wait ARQ (S&W ARQ) for a terrestrial wireless channel. Selective ARQ which is used in Seaweb system for UWASNs is another ARQ method. We show the effects and differences of the network performances by using S&W ARQ and Selective ARQ in our simulator.

Underwater Sensor Networks have received growing interests recently, which cam monitor and collect underwater environments. However, it is difficult to transmit data in underwater environments because of some effects such as severely limited bandwidth, high propagation delay, attenuation by water absorption, noises (due to movement of ships, wind, temperature etc...), and high bit error rate. There are two methods of FEC usage for underwater acoustic sensor networks. One is adding FEC for bit error protection at physical layer. Another is adding FEC for packet loss at application layer. In this paper, the former FEC usage is investigated. Furthermore, AFECCC, which was originally proposed for wireless terrestrial networks, is applied to underwater environments and investigated.

Underwater Sensor Network, a Sensor Network constructed in ocean, is different from terrestrial Sensor Networks in many points such as high propagation delay and narrow bandwidth. These differences may lead Underwater Sensor Networks to different performances. Therefore, this paper tries to evaluate the performances in the Underwater environment of three ad-hoc routing protocols, AODV, DSDV and DSR, by using network simulator ns-2.

Today, for market's investigations and automatic monitoring systems, crowd density estimation becomes an important research topic. There are existing methods like the camera based crowd density estimation or the RFID based counting systems. However, these methods need deployment cost and enough spaces to set the sensors. To solve these problems, there is a estimation system using RSSI (Received Signal Strength Indicator) and LQI (Link Quality Indicator) in wireless sensor networks. RSSI is a signal strength obtained by monitoring the packets that are communicated between nodes. We assume that, as the number of people increase, RSSI value is violated and decreases. However, this method applies only to networks that one pair nodes communicate with each other. It's not practicable for many networks. In this paper, we therefore extend a existing method to multihop networks where two or more nodes access to a base station and cover wider area. Finally, we evaluate validity of our proposal and develop an application with wireless sensors.

Recently, Underwater Sensor Network has attracted attention as a sensing method of environmental sea information. However, Underwater Sensor Network uses acoustic wave for data transmission which is very slow causes heavy transmission delay, and is vulnerable to acoustic noises in a sea. Therefore, it is very difficult to implement efficient routing and data transmission by the existing Ad-hoc routing protocols. In this paper, we propose an On-demand routing method adapted to Underwater Sensor Network, and evaluate its performance by a network simulator.

SIFT is known as local features that has robustness against rotation, scale and illumination changes of the images. However, in matching by the original SIFT, there is a problem that the keypoints on edges often become false recognition. We had therefore presented a method for alleviating this problem by apply Bilateral filter instead of Gaussian filter that is used for smoothing process in SIFT. Bilateral filter does not decide a weight only by using distances between pixels, but also considers brightness differences of pixels, and in the location where a large pixel difference is observed, it keeps edges by reducing the weight. By using this method, the amount of information of image differences increases and improvement of correspondence accuracy is expected. However, there is a problem from the viewpoint of computational complexity. Therefore, we try to develop faster implementation of the proposal by using GPGPU. We accelerated the filtering process by CUDA which causes heavy increase in computation time, and confirm its faster processing.

Recently, the performance of Internet has been improved, and the video coding technologies have also evolved. As a result, many various streaming services have been provided. Especially, many people have had a chance to enjoy these services using their mobile phones such as iPhone. Therefore, we need to focus on specific researches on video streaming over wireless networks. In this paper, we implement adaptive rate control mechanisms in VIC (Video Conferencing Tool), and evaluate their performances using Linux notebooks. We observe the effectiveness of this implementation, the fairness evaluation, and the received image evaluation when competing with a TCP-Reno flow over wired and wireless combined networks.

In Bag-of-Keypoints method used in generic object recognition, we ordinarily derive SIFT features over the whole regions of images. However, in some categories, information of background regions is not effective for recognition. Therefore, in this paper, we investigate the effect of object extraction on the recognition rate of Bag-of-Keypoints. Firstly, by using ground truth data of foreground objects, we found that classification rate can be improved by 7% in average and region extraction is effective for some categories but not for other categories. Secondly, we propose a method to extract object regions automatically by using visual attention model prior to Bag-of-Keypoints. Unfortunately, this method shows only 1.5% improvement mainly due to categories which need background regions. Finally, we propose an adaptive method which selects foreground regions or whole image for SIFT feature description according to categories. Experimental results show that above 7% improvement is furthermore expected against other methods described above.

It is important to evaluate performance of various applications over several developing network environments. One way to solve this problem is to use a network emulator which emulates an assumed virtual environment in real-time. Although data communication widely uses wireless network, there are few handy emulators which can emulate wireless environments. Recently, the network emulators, called "Scenargie" and "EXata", which emulate wireless network without any specific kernel modifications or hardware requirements, were shown up. Emulation experiments in this paper use Skype as an example of operational application. Performances of Skype over wireless network are avaluated via the wireless emulator. These evaluation results reveal how Skype adapts to the changes of network status.

The presence of free-riding peers has been known as a serious threat in the peer-to-peer networks as it may lead to the degradation of overall service quality. In this paper, we study a method to construct the push-based multicast overlay in a contribution-aware manner for P2P streaming. The idea is to construct the tree-shaped core overlay and define a set of auxiliary links to improve its resiliency. The auxiliary connections, which are constructed based on the service capacity and contribution, also alleviate the peers to deliver data using a multi-parent streaming approach. Simulations through ns2 demonstrate the efficiency of our solution.

Recently, the color depth is beginning to shift from 24 bit-per-pixel to above, for example 30 bit-per-pixel that is able to represent more color levels. Accordingly, some papers had been reported which tried to expand existing 24bpp image resources to more color levels. In this paper, we examine a method that extends color depth of images by restoring lower bits of pixel values according to interpolated color histogram, carry out experiments of two cases, from 18bpp to 24bpp and from 24bpp to 30bpp, and evaluate our method.

In this paper, we introduce a hybrid approach for overlay construction in an application-layer multicast and its application to a distributed data center management. We combine the good features of a tree-based structure and a mesh-based data delivery to form ToMo, a two-layer hybrid overlay. Different from other approaches, the overlay is constructed in the fashion of a mesh layer over a tree layer. This structure helps to reduce the number of replicated packets at a source by allocating the source to multicast each packet to a specific group of child peers only. We also consider a simple application of our overlay structure to the failover for data centers. That is, the servers of distributed data centers are connected by a mesh, and the clients are hanging on the servers in a tree structure. Upon failure, the server can be switched over to a redundant server follows the defined mesh routes automatically. Simulations through ns2 demonstrate the efficiency of this solution.

One of the typical methods to extract the features of images is SIFT. SIFT has advantages of robustness in rotation, scale change and illumination change of the images. However, in matching by the original SIFT, there is a problem that the keypoints on edges often become false recognition. This paper therefore presents a method for decreasing this problem by apply Bilateral filter instead of Gaussian filter that is used for smoothing process in SIFT. Bilateral filter does not decide a weight by only using distances between pixels, but also considers brightness differences of pixels, and in the location where a large pixel difference is observed, it keeps edges by reducing the weight. By using this method, the amount of information of image differences increases and improvement of correspondence accuracy is expected. The experiments investigate different characteristics of our proposal to the original SIFT; differences of smoothed images, feature point extraction and SIFT features. In addition, they also verify performance improvement of our proposal in searching corresponding points.

In this paper, our goal is automatic segmentation of object regions which is processed before recognition in Bag-of-Keypoints. Region segmentation, such as Graph Cuts, is useful for object recognition. Objects tend to exist at the center of image, so we give seeds for a center and four corners of an image and apply the first-stage graph cuts to segment along edges. We get a color histogram from the first segmentation results and assign probabilities as the t-link of the graph for the second-stage graph cuts. Bag-of-Keypoints is then applied to the second segmentation results. We demonstrate that we can obtain better results than the case of using only Bag-of-Keypoints method.

To improve driving comfort and provide entertainment services, vehicular communication networks (VCDNs) have appeared as an emerging solution, which consists of road-side units (RSUs) and on-board units (OBUs) to distribute multimedia contents. However, as most of OBUs always request the stored contents in the RSUs, how to update the contents in these RSUs when the original changes at its original servers has become an important issue to be dealt with. This paper proposes a novel method to resolve the above problem. Firstly, based on the characteristics of peers and geographical information, we decide which replica of which content in RSUs should be updated when its original changes. Secondly, by comparing the delivery cost of wired and wireless transmission, we decide whether the updated content should be delivered from a fixed peer or other mobile peers. Lastly, the detailed algorithm is presented and summarized.

It is difficult to charge battery of sensor nodes, considering the deployed places and the number of nodes, in wireless sensor networks. Therefore, it is desirable to collect the sensed data to base stations as energy-efficiently as possible under the energy consumption limitation. LEACH is a self-organized clustering based protocol, which is able to save the node energy and distribute energy consumption among the sensors in the network. In LEACH, by introducing clusters, local cluster-heads (CH) send the base station the aggregated data which is gathered from cluster-members. LEACH also uses randomized rotation of CH which consumes large energy. However, this protocol suffers from next two issues. First, CH selection doesn't consider residual energy of sensors. Second, cluster size is unbalanced among the clusters in the network. In this paper, we therefore propose a clustering method by improving LEACH's clustering mechanism. As a result, network lifetime of the proposal is extended to realize low-energy consumption and better load balance is achieved than LEACH.

In recent years, wireless terminals have grown widely and diversely along with evolution of video coding technologies. Accordingly, many people get a chance to enjoy video streaming services anywhere and easily. Therefore, we need to focus on specific researches on video streaming over wireless networks. In this paper, we carry out video delivery experiments using VIC (Videoconferencing Tool), which has been a reference software of video streaming since the mbone experiment, over wired and wireless combined networks. We also try various TCP protocols (TCP-Reno, CUBIC, TCP-Vegas, TCP-Fusion, and its improved version) using iperf on the shared network. We observe receiving rate of VIC's RTP packets and various properties (e.g. cwnd and RTT) of competing TCP packets.

In recent years, various TCP congestion control mechanisms have been proposed because Standard TCP(e.g. TCP-Reno) is not efficient in high bandwidth and delay networks. Today, CUBIC-TCP, which is said to have scalability and stability, is implemented and used on Linux. Our laboratory has also proposed TCP-Fusion which has hybrid congestion control mechanism and its improved version (Proposal) which achieves RTT fairness. In this paper, we evaluate the performances of CUBIC-TCP, TCP-Fusion and our new proposal on real machines. The results show Hybrid-TCPs have robustness for packet loss and high friendliness to TCP-Reno and CUBIC-TCP has good RTT-fairness as well as our proposal.

Along with downsizing and cost reduction of wireless sensor nodes, researches for wireless sensor networks have become popular. In the wireless sensor networks, since sensor nodes are supplied by battery in many case, we need energy efficient communication methods. Examples of the conventional methods are "LEACH" based on TDMA "SMAC" based on CSMA and "ZMAC" based on both TDMA and CSMA. We propose a transmission method which combines TDMA and CSMA like ZMAC. ZMAC changes CSMA to TDMA when nodes sense collisions. On the other hand, our proposal does not change CSMA to TDMA, but it divides the communication areas into TDMA area where nodes transmit only by TDMA and CSMA area where nodes transmit only by CSMA.

TCP in wireless multi-hop ad hoc networks receives various influences like a peculiar loss in wireless environment, hidden terminal problem, exposed terminal problem and so on. Therefore, the performance of TCP decreases. In addition, conventional TCP protocols decrease the performance in wireless multi-hop ad hoc networks. Because conventional TCP protocols are designed for wired networks, the window size raises aggressively. Therefore, it causes the contention in MAC layer, packet collisions and the retransmission timeout. As a result, it leads to throughput decreases. Moreover, the throughput of TCP-Vegas is known to be about 30%〜70% higher than TCP-Reno. We propose a congestion control method based on TCP-Vegas for wireless multi-hop ad hoc networks. We compare the proposal method with the conventional methods(Reno, Vegas, Ada Vegas, Vegas-W).

This paper presents performance improvement of TCP-Vegas combining the 802.11e priority control and window control over Wireless LAN. In TCP uplink connection over Wireless LAN, the problem is indicated that unfairness among TCP flows occurs because downlink packets stay in access point (AP). In addition, when we observe behaviors of TCP-Vegas, RTT increases by the packets staying in AP and total throughput (bandwidth utilization efficiency) can't get enough. This paper pays attention on this point, and we propose a method of TCP-Vegas assuming the 802.11e priority control and mild window control, which presents better fairness among TCP flows and throughput improvement.

In this paper, we introduce a hybrid approach for overlay construction and data delivery in an application-layer multicast. We combine the strong points of a tree-based structure and a mesh-based data delivery to form ToMo, a two-layer hybrid overlay. We try to reduce the number of replicated packets at a source, and reduce an effect when slow connection peers are located near the source. The overlay is constructed in the fashion of a mesh layer over a tree layer. This structure allocates the source to multicast each piece of the packet to a specific group of child peers only. Different from other approaches, we employ only push-based data delivery in order to minimize the latency. The redundancy is avoided by defining a set of well-organized mesh connections. Furthermore, in our approach, the isolated peers affected by parent departure are not facing data loss during the rejoin process since they still receive data from their neighbors via mesh connections. Simulations through ns2 demonstrate the efficiency of this solution.

The hybrid Application Layer Multicast (ALM) has been shown its efficiency by leveraging the conventionally main structures of application layer multicast, tree-based and mesh-based. However, how to select the proper node to construct the overlay and how to establish the connection between any two nodes are still unsolved. Therefore, this paper is to design a novel construction algorithm for the hybrid AIM to resolve the above two issues. Firstly, by carrying out the analysis of nodes&apos; characteristics, all nodes are divided into groups and a node priority is proposed to select the super node within each group. Secondly, by using the selected super node, all nodes are hierarchically controlled and different kinds of connections are carried out in the AIM, where the connection between super node and other nodes is set to be a tree to enhance the efficient utilization of network resource while the connection between other normal nodes is decided to be a mesh to reduce the overhead. Simulation results show that the proposal outperforms other conventional methods.

With the development of broadband technologies, more and more multimedia contents such as video or bulk file are being distributed over the Internet. However, an important issue to be resolved is how to construct an efficient overlay to provide users with the acceptable throughput and delay, while receiving multimedia contents. Therefore, this paper designs a tree construction algorithm by distributing the layered steaming over the ALM in order to improve both throughput and user delay. Firstly, to improve the throughput, based on theory analysis, we define a layered degree, out/in-degree and the corresponding constraints to manage the layered streaming and nodes. Secondly, a novel method, called Bi-cast, is proposed to reduce user delay during data-transmission. Thirdly, by using the defined degrees and the Bi-cast, we present a tree construction algorithm. Both the simulation and the implementation are carried out, and the results show that our proposal can obtain better performance than other conventional methods.

This paper presents a hybrid congestion control algorithm supporting throughput efficiency, TCP friendliness and RTT fairness, in order to provide highly smart multimedia streaming over the Internet. Recent studies on TCP and TFRC are evolving to achieve better throughput efficiency and RTT fairness. Examples are Hybrid TCP congestion control, VTP and TCP-Libra, where the former two improve throughput efficiency of TCP and TFRC, respectively, and the last one improves RTT fairness of TCP. This paper integrates these ideas and propose a new congestion control algorithm, called Hybrid TFRC, for multimedia streaming. Experiments are carried out and effectiveness of our proposal is verified. ©2010 IEEE.

Service capacity and contribution of the peers have been proven to be the important factors for selecting a serving peer in the P2P overlay networks. In this paper, we study how to design a multi-parent overlay network in a contribution aware manner. Our idea is to utilize the outbound bandwidth of all overlay nodes efficiently. Different from other multisource streaming approaches, we employ push-based data delivery in order to minimize the latency. The core of our overlay is conceived in a tree structure rooted at a source. The auxiliary connections with redundancy avoidance capability are proposed to increase the resilience of the tree. Furthermore, we introduce an overlay joining algorithm where the parent whose degrees are exhausted responses the new peer with a list of potential peers having residual degrees so that the new peer can join the overlay with minimal requests. Simulations through ns2 demonstrate the efficiency of our solution. ©2010 IEEE.

With the development of recent vehicular communication technologies, distributing multimedia contents in the vehicular communication networks (VCNs) has become more and more popular, to provide conveniences and entertainment services during the time of driving. However, as multimedia contents are changed and updated dynamically, how to keep the consistency between the original and these replicas in VCNs is very important. Therefore, this paper designs a novel algorithm to control the consistency for the VCNs. In our proposal, after the analyses of the status of road-side units, on-board units and local geographical information, we divide all replicas into different groups, where one is necessary for update and the other are not. Then, we compare the cost to update replicas by using wireless and wired connection, and propose a method to make selection between them. The performance of our proposal is tested by simulation experiments. And the results show that our method can obtain the low delay successfully.

This paper proposes a new TCP congestion control supporting loss-fairness, which also contributes to improving RTT fairness. It has been well known that smaller RTT flows grab more bandwidth than larger ones when they compete on the shared link. This is called RTT unfairness and is caused by congestion control mechanism of the current TCP. To solve this problem, many protocols have been proposed. TCP-Libra is one of them, which formulates its window increase rate as a function of RTT and brings the same throughput irrespective of different RTT values. However, we indicate TCP-Libra's formulation holds only when the congestion window is halved upon packet losses similar to TCP-Reno. When we apply different (smarter) window control schemes, the formulation has to be modified accordingly. Therefore, in this paper, we consider a hybrid congestion control supporting RTT fairness which switches two modes according to buffering state at the bottleneck link. In decrement phase upon packet loss, it decreases window size to clear the buffering packets instead of halving the window size. Furthermore, we introduce the concept of loss-fairness for our window increase formulation, which is equivalent to RTTfairness. Experiments are carried out to validate the proposed method by both implementation experiments and simulations, and much better performances are achieved against conventional methods. ©2010 IEEE.

In this paper, we introduce a new approach for overlay construction in application layer multicast. We combine the good features of tree structure and mesh data delivery to build a two-layer hybrid multicast overlay. The overlay is constructed in a fashion of mesh layer over tree layer. This structure, in an un-optimized overlay, helps to alleviate content bottleneck causing by slow connection peers. Moreover, the peers affected by parent departure are not facing data outage during rejoin process since it still receives data through mesh. Simulations using ns2 demonstrate the efficiency of our solution. Laboratory

Recently, a demand for generic image recognition is increasing due to explosive increase of digital images. In this paper, we carried out an experiment by using bag-of-keypoints method (Bok method) proposed by G. Csurka as a technique of image recognition. In this experiment, we tried with an assumption that, by removing an area that seemed to be unnecessary for recognition by applying seam-carving to both the learning images and the test images, learning efficiency and recognition rate could be improved. As a result, the recognition rate can be improved by about 4.5% compared with the case where nothing is processed beforehand.

Recently, construction of the system that retrieves similar music from the large-scale music data base is expected. As a metric of the music similarity, estimation of the emotion feeling in listening to music has been studied. As a typical method, music emotion classification to four emotion clusters defined by Thayer's musical psychology experiment had been proposed. However, categorized emotions are too ambiguous for music information retrieval. Therefore, in this paper, we calculate arousal and valence values of Thayer's model, and show distribution of music emotions. Furthermore, overlapping clustering is applied to improve accuracy.

Picture quality assessment tries to measure deterioration of the image. There are subjective assessment and objective assessment. The latter saves cumbersome tasks of the subjective assessment, and is very important for the studies of image compression. However, it has been pointed out that PSNR (Peak signal to Noise Ratio), which is widely used as an objective quality measure, has not sufficient correlation with the subjective quality assessment results. For example, there are blocking artifacts due to DCT and flicker noises caused by temporal changes of image luminance and color. SSIM (Structual SIMilarity) has been recently proposed as an objective quality measure, but it cannot reflect temporally oscillating flicker noises. This paper therefore focuses on flicker noises, and tries to improve correlations of PSNR and SSIM to subjective quality assessment results by incorporating flickering effect into the measures.

With the rapid development of wireless communication technologies, a mobile contents delivery network (M-CDN) has been proposed and becomes one of the most important approaches to deliver contents to mobile terminals. However, whereby mobile content distribution systems can manage dynamically changing files, an important issue to be resolved is consistency management, which means the cached replicas on different mobile sites must be updated if the originals change. This paper is to talk about the consistency control for mobile contents delivery network. And a novel consistency control algorithm is proposed to enhance the efficient utilization of network resource and to support the client mobility. Finally simulation results show that the proposal outperforms other conventional methods.

On attempting live video streaming over wireless networks, an appropriate congestion control which provides robustness against wireless losses is imperative. This is because the congestion controls designed for wired networks consider all the packet losses as congestion indication and reduce their sending rate unreasonably. VTP (Video Transport Protocol) is a congestion control designed for wireless video transmission by using AR (Achieved Rate) which is the actual receiving rate. In this paper, we investigate that AR is not measured correctly when the encoding delay is larger than the target frame rate. And we propose another method correcting AR measurement with the cooperation between encoder and congestion control. Moreover, we confirmed the effectiveness of our proposal in actual wireless environment.

In this paper, we propose and evaluate methods to visualize the acoustic feature space of the content-based music information retrieval (MIR) system. Our methods map all songs in the music collection, and enable users to intuitively comprehend the acoustic feature space by the location of songs, thus is expected to reduce retrieval time. In this paper, we have developed a prototype visualization MIR system, and proved the effectiveness of our proposal, in situations where the users search for songs of various genres from a large-scaled music collection.

This paper presents a novel key estimation method of sound sources based on the music theory known as "circle of fifths". We firstly overview music theory and formulate the musical key analysis by vector operations. In detail, we separate music sources into small pieces and calculate FFT-based chroma vectors. They are converted to tonality vectors and COF (circle-of-fifth) vectors are calculated from the tonality vectors, which are mapped onto the circle of fifths coordinate. As a result, each music source can be represented by traces of COF vectors, which usually stay inside a single key region on the circle of fifths. Finally, HMM is applied to the traces of COF vectors in order to detect keys and their boundaries. Experiments using music databases are also carried out.

In order to achieve highly accurate content-based music information retrieval (MIR), it is necessary to compensate the various bit rates of encoded songs which are stored in the music collection, since the bit rate differences are expected to apply a negative effect to content-based MIR results. In this paper, we examine how the bit rate differences affect MIR results, propose methods to normalize MFCC features extracted from encoded files with various bit rates, and show their effects to stabilize MIR results.

With the growing use of dynamic content by mobile content distribution systems, how to manage dynamically changing files has become an important issue, since the cached replicas on different mobile sites must be updated if the originals have been changed. Therefore, this paper proposes a priority based selection method to enhance the efficient utilization of network resource and support the client mobility for Mobile Contents Delivery Network (M-CDN). On one hand, a consistency priority is calculated by analyzing the characteristics of mobile surrogates. If a given content which has been changed on its original node, only the replicas with the high consistency priority instead of all replicas are updated. On the other hand, an update priority is also proposed. If a replica has been selected for update, the latest version will be sent from the site decided by the update consistency. Simulation results show that the proposed new approach outperforms other conventional methods. © 2009 IEEE.

This paper presents TCP-Fusion supporting RTT (Round Trip Time) fairness in addition to throughput efficiency and friendliness to TCP-Reno. When multiple TCP flows having different RTT values compete, more bandwidth is unfairly allocated to the flow having smaller RTT. This means that a user with longer RTT may not be able to obtain sufficient bandwidth by the current methods. On the other hand, recent studies on the TCP congestion control to achieve RTT fairness and throughput efficiency are evolving actively. An example for RTT fairness is TCP-Libra and an example for throughput efficiency is Hybrid TCP congestion control. This paper focuses on Hybrid TCP (exploiting residual link capacity when TCP-Reno drops its rate) and improves its RTT fairness by incorporating the ideas of TCP-Libra (congestion window increase in proportion to square of RTT) to its loss mode. Experiments are carried out to validate the proposed method and much better performances in RTT fairness and throughput are provided against conventional methods.

In this paper, we introduce a new mesh-based approach for data delivery which is organized over multiple tree-shaped core routes. Given that both tree and mesh approaches for multicast overlay construction have their own strong points, we simply combine them. During session, the mesh connections are formed dynamically to facilitate the data delivery on tree-shaped core routes which may experience bandwidth bottlenecks. Unlike existing mesh-based approaches, our proposal employs push-based data delivery along mesh connection. A well-organized mesh construction scheme is defined to avoid data redundancies. We evaluated the proposal through ns-2 simulator. The simulation results demonstrate that our approach can provide higher average received quality and has acceptable control overhead when compared to a single tree method. We also show that, over a static overlay, the push-based data delivery on mesh can provide the received quality close to supposed pull-based data delivery method with less latency. Moreover, it has lower control overhead than the pull-based method when the peer number is large.

Sensor Network has been a hot research topic for the past decade and has moved its' phase into using multimedia sensors such as cameras and microphones. Combining many types of sensor data will lead to more accurate and precise information of the environment. However the use of the sensor network data is still limited to certain circumstances. Thus we propose the deployment of sensor data as web services. In order to unify different types of sensor data and also to support heterogeneous client applications, we used Resource Oriented Architecture.

Sensor Network has been a hot research topic for the past decade and has moved its' phase into using multimedia sensors such as cameras and microphones. Combining many types of sensor data will lead to more accurate and precise information of the environment. However the use of the sensor network data is still limited to certain circumstances. Thus we propose the deployment of sensor data as web services. In order to unify different types of sensor data and also to support heterogeneous client applications, we used Resource Oriented Architecture.

Sensor Network has been a hot research topic for the past decade and has moved its' phase into using multimedia sensors such as cameras and microphones. Combining many types of sensor data will lead to more accurate and precise information of the environment. However the use of the sensor network data is still limited to certain circumstances. Thus we propose the deployment of sensor data as web services. In order to unify different types of sensor data and also to support heterogeneous client applications, we used Resource Oriented Architecture.

This paper is to design an integrated consistency-control algorithm for mobile contents delivery network (M-CDN) to enhance the efficient utilization of network resource and support the client mobility. Firstly, by carrying out an analysis of mobile surrogates&#039; characteristics, for a given content which has been changed on its original node, only a limited number of its replicas instead of all replicas are updated. Then, if a replica has been selected for update, the latest version will be sent from an algorithm-decided site instead of from its original server.

In this paper, we introduce a new mesh-based approach for overlay construction which is organized over multiple tree-shaped core routes. We evaluate the proposal through ns-2 simulator. The results demonstrate that our approach can provide higher in average received quality and has acceptable data delivery delay when compared with a fundamental tree-based method.

This paper presents a novel key estimation method of sound sources based on the music theory known as "circle of fifth". We firstly overview music theory and formulate the musical key analysis by vector operations. In detail, we separate music sources into small pieces and calculate FFT-based chroma vectors. They are converted to tonality vectors and COF (circle-of-fifth) vectors are calculated from the tonality vectors. Finally they are mapped onto the circle of fifth coordinate. As a result, each music source can be represented by traces of COF vectors, which usually stay inside a single key region on the circle of fifth. We can detect keys and their boundaries by analyzing COF vectors with HMM.

In this research, we implement an application prototype to detect human 3D location which calibrates a low resolution single stationary monitoring camera, detects human in the camera image by using Haar-like features of human body in real-time, estimates 3D location by using geometric conversion of under foot position, and tracks its movement in temporal direction.

This paper presents comparison results of denoising methods for JPEG compressed images, especially focusing on reduction of quantization errors and blocking artifacts. We overview typical conventional methods like POCS and TV and provide their performance comparison results using actual still images. In addition, we consider combination of the above methods with the technique we had proposed for quality improvement of the Motion JPEG image sequences by using motion estimation and pixel shift, and report some experiment results.

Recently, researches on new image encoding techniques such as MDC (Multiple Description Coding) and DSC (Distributed Source Coding) are advanced. We show that PSNR values are improved by synthesizing multiple compressed images based on the MDC technique. Furthermore, in this paper, we propose a new synthetic technique of compressed images and verify an effect of image quality improvement of the proposal method by subjective evaluation experiments.

This paper presents similar music retrieval that uses acoustic features and lyrics features of musical pieces. In typical content-based music retrieval, the features of music pieces are expressed by vectors, and music pieces are retrieved based on their similarities. When generating the features vectors, using "features that influence impression that users receive from music" as elements improves retrieval accuracy. Therefore, in this paper, we use lyrics features in addition to acoustic features that are extracted from music signals. Furthermore, we estimate the efficacy of this music retrieval method by a subjective evaluation experiment.

With the development whereby mobile content distribution systems can manage dynamically changing files, an important issue to be resolved is consistency management, which means the cached replicas on different mobile sites must be updated if the originals change. This paper is to design an integrated consistency-control algorithm for mobile contents delivery network (M-CDN) to enhance the efficient utilization of network resource and support the client mobility. Firstly, by carrying out an analysis of mobile surrogates' characteristics, for a given content which has been changed on its original node, only a limited number of its replicas instead of all replicas are updated. Secondly, if a replica has been selected for update, the latest version will be sent from an algorithm-decided site instead of from its original server. Simulation results show that the proposal outperforms other conventional methods.

Recently, Underwater Sensor Network has attracted attention as a new network in the sea. However, the data transmission in this undersea environment suffers from various influences such as bandwidth usage limitation, surrounding noise, and huge acoustic transmission delays. Therefore, communication itself is a difficult problem compared with on-the-ground wireless environment. Moreover, an efficient communication cannot be done by the conventional protocols. The congestion window size doesn't grow up sufficiently due to large transmission delays when conventional transport-layer protocols are applied in the underwater environment. Therefore, this paper investigates a usage of TCP Hybla suitable in the underwater environment, which was originally designed for satellite channels. We carry out performance comparisons with the conventional TCPs (for instance, Reno) and also evaluate the influence of ad-hoc routing protocols on the transport performance

Recently, with the spread of real-time services, new congestion control methods for real-time communication in IP network has been proposed, like TFRC (TCP-Friendly Rate Control) and VTP (Video Transport Protocol). These methods are designed based on the AIMD algorithm and the equilibrium rate of TCP Reno. However, there is a problem that Reno sessions cannot receive the same share of bandwidth, when their RTTs (round trip delay) are very different among sessions. An evaluation result using a computer simulation shows that even TFRC and VTP based on Reno have a similar problem. In this paper, we especially pay our attention to VTP, and propose a congestion control method supporting fair-share bandwidth, by setting an increase coefficient in proportion to RTT. We also evaluate it by computer simulations and show the effectiveness.

In ubiquitous networks, along with evolution of downsizing and power-saving capabilities of sensor chips and wireless networking devices thanks to rapid developments in semiconductors and wireless networking technologies, wireless sensor networks have come into practical use. Furthermore, recently, various trials to use moving wireless sensor nodes for mobile sensor networks have been researched actively. Especially in this paper, we pay our attention to the localization problem. We use publicly available sensor nodes and mobile robots as a platform, and we report experimental results in outdoors and indoors by using the SLAM (Simultaneous Localization and Mapping) for localization estimation.

We performed comparative experiments of adaptive rate controls for video transmission over wireless ad-hoc network. TFRC (TCP Friendly Rate Control), which is one of the effective rate control methods for video streaming over wired network, does not function effectively over ad-hoc network because of non-congestion packet losses. On the other hand, VTP (Video Transport Protocol) was proposed against this problem with LDA (Loss Differentiation Algorithm) and AR (Achieved Rate). We considered the suitable method for video streaming in contention with these protocols over ad-hoc network.

It is possible to transmit contents to a great many people without increasing server loads by using Peer-to-Peer(P2P). There is a distributed hush table (DHT) scheme as one of the P2P technology. However, there are problems in the DHT which forces heavy load to a node which maintains contents of many access requests regardless of its processing capability. Therefore, we discuss content replication methods onto DHT's ID spaces based on the number of accesses in order to solve the load balancing problems and clarify the effectiveness of our proposal by simulation evaluations.

Recently, as the Internet is evolving to high-speed networks, a problem has been actualized that the current TCP (mostly TCP-Reno and TCP-SACK) cannot fully utilize available bandwidth in such high-speed networks with large bandwidth-delay product. To address this issue, new loss-based congestion controls such as Highspeed-TCP and Scalable-TCP and delay-based controls such as FAST-TCP have been proposed. Furthermore, new approaches called Hybrid-TCP combining the loss-based control with the delay-based control have been proposed. In this paper, we present performance improvement methods of YeAH-TCP and Compound-TCP which are typical examples of the Hybid-TCP. In these methods, we try to improve their throughput efficiencies in the high-speed environment by modifying their congestion window controls, and also investigate negative influences caused by the proposed methods.

In order to realize highly accurate content-based music information retrieval (MR), it is necessary to compensate with the various bit rates of the songs which are included in the music collection, because the bit rate differences are expected to apply a negative effect to MIR results. In this paper, we propose methods to normalize MFCC features extracted from MP3 files with various bit rates, and analyze their effects to content-based MIR.

This paper presents a denoising method of intra-coded pictures using motion estimation and pixel shift. Firstly, we show that pixel-aligned mixture of distorted images which are spatially shifted and differently encoded brings reduction of quantization errors. We show that this effect can be formulated as a special case of Wiener-Hopf equation and independence of quantization errors affects the performance. We then consider its application to denoising of intra-coded pictures by using motion estimation and pixel shift. Experiments using actual image sequences verify that motion estimation is effective in moving regions, pixel shift is effective in static regions and favorable PSNR gains are achieved.

With the development whereby mobile content distribution systems can manage dynamically changing files, an important issue to be resolved is consistency management, which means the cached replicas on different mobile sites must be updated if the originals change. This paper is to design an integrated consistency-control algorithm for mobile contents delivery network (M-CDN) to enhance the efficient utilization of network resource and support the client mobility. Firstly, by carrying out an analysis of mobile surrogates' characteristics, for a given content which has been changed on its original node, only a limited number of its replicas instead of all replicas are updated. Secondly, if a replica has been selected for update, the latest version will be sent from an algorithm-decided site instead of from its original server. Simulation results show that the proposal outperforms other conventional methods.

This paper presents simple analytical models of hybrid TCP congestion controls, which switch loss-based mode and delay-based mode adaptively, and tries their performance tuning. We firstly present ideal behavior models of three kinds of TCP congestion controls (loss-based, delay-based and hybrid). We then give abstracted models of the actual hybrid TCP s and consider their performance tuning. Finally, experiments validate analytical expectations and effectiveness of the hybrid TCP.

This paper presents a new video coding method which improves PSNR of decoded pictures by using additional information. Firstly, we show a pixel shift effect which brings PSNR improvement by composing multiple compressed images extracted from an original image and, as an application of the pixel shift effect, we then propose a video coding method which sends higher precision motion vectors to a decoder as additional information. We especially focus on how to select effective motion vectors at an encoder and how to decide mixture of decoded pictures for improving PSNR. Finally, we show experimental results that present improving R-D characteristics of H.263+.

This paper presents a new video coding method which improves PSNR of decoded pictures by using additional information. Firstly, we show a pixel shift effect which brings PSNR improvement by composing multiple compressed images extracted from an original image and, as an application of the pixel shift effect, we then propose a video coding method which sends higher precision motion vectors to a decoder as additional information. We especially focus on how to select effective motion vectors at an encoder and how to decide mixture of decoded pictures for improving PSNR. Finally, we show experimental results that present improving R-D characteristics of H.263+.

This paper presents new video coding methods which improve PSNR of decoded pictures by mixing corresponding pixels of neighboring frames based on the distributed video coding framework. Firstly, we show that we can statistically reduce quantization errors by mixing distorted pixels in the same image which are independently compressed. Secondly, based on this principle, we propose new video coding methods which exploit motion estimation and intentional pixel shift mainly at a decoder side. Finally, we show several experimental results using Motion JPEG and H.263 extensions to actual image sequences and validate effectiveness of our proposed methods.

This paper presents mathematical framework for hybrid TCP congestion control. Recently, several hybrid TCP congestion control methods have been proposed, in which loss-based features and delay-based features of the TCP congestion control are combined in a smart way. However, since its analytical performance has not yet been clarified sufficiently, we try to develop simple mathematical framework to evaluate and tune its performance. The results prove efficiency and robustness of the hybrid congestion control when random packet losses frequently happen, and also prove its good friendliness to TCP-Reno when stable channels are available. We furthermore present delay-based TCP friendly rate control, which provides better throughput performance than classical TFRCs.

Recently, as congestion control methods for realtime communication in IP network, many researchers have an interest in TFRC (TCP-Friendly Rate Control) and DCCP (Datagram Congestion Control Protocol). In addition, rapid increase of real-time services for mobile terminals is expected together with the spread of FMC and WiMAX. Therefore, we pay our attention to handover characteristics in heterogeneous networks. Especially, when the delay is too large between a mobile terminal and an AP (Access Point) after handover, unnecessary rate reduction and timeout will happen. Since a TFRC receiver cannot immediately generate feedback packets adapted to the heterogeneous environment after handover, it might increase sending rate too large and negatively impacts the targeted network. In this paper, we therefore propose a ACK management method that can reduce unnecessary rate reduction, and furthermore, we also propose a continuous ACK management method and a packet loss rate notification method that can adequately applied to the heterogeneous environment after handover.

We performed a comparative experiment of some rate controlling protocols for video transmission over wireless ad-hoc networks. One of the protocols is TFRC (TCP Friendly Rate Control [1]) which is an effective protocol for video streaming on wired networks. But it is also known for its ineffectiveness on wireless networks with random losses. By comparing these protocols, we considered the suitable protocol for video transfers on wireless ad-hoc networks.

Recently, as congestion control methods for realtime communication in IP network, many researchers have an interest in TFRC (TCP-Friendly Rate Control) and DCCP (Datagram Congestion Control Protocol). In addition, rapid increase of real-time services for mobile terminals is expected together with the spread of FMC and WiMAX. Therefore, we pay our attention to handover characteristics in heterogeneous networks. Especially, when the delay is too large between a mobile terminal and an AP (Access Point) after handover, unnecessary rate reduction and timeout will happen. Since a TFRC receiver cannot immediately generate feedback packets adapted to the heterogeneous environment after handover, it might increase sending rate too large and negatively impacts the targeted network. In this paper, we therefore propose a ACK management method that can reduce unnecessary rate reduction, and furthermore, we also propose a continuous ACK management method and a packet loss rate notification method that can adequately applied to the heterogeneous environment after handover.

We performed a comparative experiment of some rate controlling protocols for video transmission over wireless ad-hoc networks. One of the protocols is TFRC (TCP Friendly Rate Control [1]) which is an effective protocol for video streaming on wired networks. But it is also known for its ineffectiveness on wireless networks with random losses. By comparing these protocols, we considered the suitable protocol for video transfers on wireless ad-hoc networks.

In this research we propose a system design of visualized applications, in which a user carries a mobile terminal enabling visualization of environmental information in ubiquitous networks. We use Web Services for the system design in order to simplify software development by using standardized programming interfaces. Core technologies of our visualized applications are marker detection in actual space and interface enhancement by using augmented reality. We also consider a human distribution detection system, in which sensor fusion technologies are utilized.

This paper is to design an integrated consistency-control algorithm for mobile contents delivery network (M-CDN) to enhance the efficient utilization of network resource and support the client mobility. Firstly, by carrying out an analysis of mobile surrogates&#039; characteristics, for a given content which has been changed on its original node, only a limited number of its replicas instead of all replicas are updated. Then, if a replica has been selected for update, the latest version will be sent from an algorithm-decided site instead of from its original server.

Along with evolution of downsizing and power-saving capabilities of sensor chips and wireless networking devices with rapid developments in semiconductors and wireless networking technologies, wireless sensor networks have come into practical use. Ubiquitous networks, which enable seamless connection of "anytime, anywhere, by anything and anyone" property, provide context-aware services, in which location services play a critical role especially. In recent years, many studies had focused on localization techniques. However, current passive localization techniques need many beacon sensors of which locations have to be preassigned and cost expensive. Therefore, this paper experiments an active and low-cost localization technique using home robots in wireless sensor networks.

Application Layer Multicast has been attracting attention in recent years as a distribution technique of multimedia streaming. One of the unavoidable problems of the ALM is that nodes leave or fail the multicast session, and it affects the session performance of nodes located in the subtree of departure nodes. In this paper, we propose an algorithm of tree construction considering node stability based on node access history, and show that the algorithm improves the problem.

This paper is to design an integrated consistency-control algorithm for mobile contents delivery network (M-CDN) to enhance the efficient utilization of network resource and support the client mobility. Firstly, by carrying out an analysis of mobile surrogates characteristics, for a given content which has been changed on its original node, only a limited number of its replicas instead of all replicas are updated. Then, if a replica has been selected for update, the latest version will be sent from an algorithm-decided site instead of from its original server.

Content delivery network improves end-user performance by replicating Web contents on a group of geographically distributed sites interconnected over the Internet. However, with the development whereby content distribution systems can manage dynamically changing files, an important issue to be resolved is consistency management, which means the cached replicas on different sites must be updated if the originals change. In this paper, based on the analytical formulation of object freshness, web access distribution and network topology, we derive a novel algorithm as follows: (1) For a given content which has been changed on its original server, only a limited number of its replicas instead of all replicas are updated. (2) After a replica has been selected for update, the latest version will be sent from an algorithm-decided site instead of from its original server. Simulation results verify that the proposed algorithm provides better consistency management than conventional methods with the reduced the old hit ratio and network traffic.

With the development whereby mobile content distribution systems can manage dynamically changing files, an important issue to be resolved is consistency management, which means the cached replicas on different mobile sites must be updated if the originals change. This paper is to design an integrated consistency-control algorithm for mobile contents delivery network (M-CDN) to enhance the efficient utilization of network resource and support the client mobility. Firstly, by carrying out an analysis of mobile surrogates&apos; characteristics, for a given content which has been changed on its original node, only a limited number of its replicas instead of all replicas are updated. Secondly, if a replica has been selected for update, the latest version will be sent from an algorithm-decided site instead of from its original server. Simulation results show that the proposal outperforms other conventional methods.

The purpose of this study is to construct a robust and stable overlay tree of ALM (Application Layer Multicast) for real-time video transmission. Firstly, we propose a proactive route maintenance which enables fast recovery of the overlay tree against node departures and failures. By forcing free node degrees for route backup, signaling overheads for route maintenance can be reduced in a scalable manner. Secondly, we improve performance of the proposed method by introducing layered video coding. Smooth layer management instead of coarse degree management contributes to reduction of the depth of overlay tree (i.e. delay) as well as to efficient bandwidth utilization. Thirdly, we introduce session records of each node into the overlay tree construction. Long-term history of user access records is expected to bring more robust overlay trees. Finally, we carried out extensive evaluation experiments. Simulations and implementations demonstrate that our methods lead to fast recovery of the overlay tree against node departures and reduction of the signaling overheads. Furthermore, introduction of layered video coding is proved to be efficient to reduce delays, to improve bandwidth utilization and to avoid severe degradation of picture quality. Final experiments show that incorporation of node stability reduces the number of nodes affected by parent node departures by promoting adequate nodes to upper layer of the ALM tree. ©2007 IEEE.

Content Delivery Overlays improves end-user performance by replicating Web contents on a group of geographically distributed sites interconnected over the Internet. However, with the development whereby overlay systems can manage dynamically changing flies, an important issue to be resolved is consistency management, which means the cached replicas on different sites must be updated if the originals change. In this paper, based on the analytical formulation of object freshness, web access distribution and network topology, we derive a novel algorithm as follows: (1) For a given content which has been changed on its original server, only a limited number of its replicas instead of all replicas are updated. (2) After a replica has been selected for update, the latest version will be sent from an algorithm-decided site instead of from its original server. Simulation results verify that the proposed algorithm provides much better consistency management than conventional methods with the reduced the old hit ratio and network traffic. © 2007 IEEE.

A robust replication algorithm is very important to retrieve Web objects from the Content delivery network (CDN), where both the network topology and Web access are changed dynamically. However, current replica strategies in CDN are to statically and repeatedly keep the complete replica of the original object on many content servers. This method is weak in adapting the dynamic network, resulting in too much consumed server space and a waste of the storage cost. It becomes more serious by replicating some large-sized objects such as streaming media, which are being distributed over the Internet,more and more. In this paper, we discuss a robust replication strategy for scalable video strew-ning in CDN to reduce user response and storage cost as follows: 1) For content servers, which layer of which video contents should be selected to replicate? 2) How to test the robustness under different network environments? Simulation results show that the proposed algorithm is robust and can provide much better performance than the conventional methods.

We propose an on-demand multipath routing algorithm in a mobile ad hoc network for video transmission and evaluate its real world performance. There have been a number of multipath routing protocols extended from AODV which is a well-known single path routing protocol. Multipath routing protocols indicate good performance in the reduction of route discovery latency and unnecessary routing packets in simulations. We show that the route establishment using source route lists provided by us [7, 16] and using received signal strength as a route metric can reduce the route discovery latency, select stable routes automatically. We evaluate this proposed method comparing with the original AODV by using eight laptop PCs.

Recently, the Internet is evolving to high-speed network, however, it has been actualized that current TCP (mostly TCP-Reno and TCP-NewReno) cannot fully utilize the available bandwidth in such as high-speed network with large bandwidth-delay product. To address this issue, a number of different approaches have been proposed. However their potential unfriendliness to TCP-Reno has hampered their widely deployment, In this paper, we propose a new congestion control algorithm of TCP. This new protocol makes use of the three useful characteristics; TCP-Reno, TCP-Vegas, and TCP-Westwood, which we call TCP-Fusion. Our proposal, by modifying these characteristics to be a scalable manner, provides a good balance among high efficiency and friendliness.

Application Layer Multicast has been attracting attention in recent years as a distribution technique of multimedia streaming. One of the unavoidable problems of the ALM is that end hosts have to reconstruct the overlay tree after a node leaves the multicast session or fails, and the algorithms to reconstruct the ALM tree quickly had been reported. In this paper, we focus on one of those algorithms, and show that the algorithm with Layered Coding improves the overall performance.

This paper proposes a training data extraction (TDE) method for user preference based music information retrieval (MIR). In conventional MIR, the biased distribution of songs in the feature space is a cause of MIR accuracy decrease. In this research, we propose a TDE method based on K-means clustering results. Experiment results prove that the proposed method solves the biased distribution problem, and improves MIR accuracy. However, there is a well-known prototype setting problem for K-means clustering. We have conducted further experiments to compare clustering algorithms, and examine the optimal clustering method for user preference based MIR.

The purpose of this study is to maintain efficient backup routes for reconstructing overlay trees quickly. In most conventional methods, after a node leaves the trees, its child nodes start searching for the new parents. In this reactive approach, it takes a lot of time to find a new parent. In this paper, we propose a proactive approach to finding a next parent as the backup route node over the overlay tree before the current parent leaves. A proactive approach allows a node to find its new parent node immediately and switch to the backup route node smoothly. In our proposal, the structure of the overlay tree using a redundant degree can decide a backup route node without so much overhead. Simulations demonstrate our proactive approach can recover from node departures 2 times faster than reactive approaches, and can construct overlay trees with lower overheads than another proactive method. Additionally we carried out experiments over actual networks and their results support the effectiveness of our approach. We confirmed that our proposal achieved better streaming quality than conventional approaches. Copyright © 2006 The Institute of Electronics, Information and Communication Engineers.

We propose an on-demand multipath routing algorithm in a mobile ad hoc network for video transmission and evaluate its real world performance in video streaming application. There have been a number of multipath routing protocols extended from AODV which is a well-known single path routing protocol. Multipath routing protocols indicate good performance in the reduction of route discovery latency and unnecessary routing packets in simulations. We show that the route establishment using source route lists provided by us (Hu and Johnson, 2002) can reduce the route discovery latency, select stable routes automatically, and work well for live video streaming without limitation of the hop count based approaches. We evaluate this proposed method compared with the original AODV by using eight laptop PCs and demonstrate live streaming experiments.

The current video switching/composition scheme employed is fundamentally subject to a variable delay of 0-1 frame at each stage due to the use of a frame synchronizer. We can fix and minimize this variable delay by controlling the video sync phase of the distributed video sources at the start of the communication session if they are driven by either the same video sync frequency or independent but very accurate sync frequencies. Furthermore, delivering the necessary phase controlling signals between the terminal and the video switching or composing point can be relatively easy because of the widely available IP networks. Based on .these ideas, we determined the performance of an experimental system we constructed that used video phase synchronization of distributed video sources. We describe the system's architecture, discuss the sync source dependency of the system, and report the delay observed in existing and proposed schemes of video composition using real time hardware experiments. The results obtained indicate that the proposed video sync controlling system is effective and has practical applications for fixing and minimizing the video switching/composition delay.