Recently, aging of the population and medical cost inflation are emerging as social issues to be solved. To address this problem, estimations of bioinformation based on Internet of Things (IoT) and machine learning get more attention to researchers. In this paper, in order to estimate R-R Interval (RRI) without using specialized and professional wearable devices, we propose a deep learning based RRI estimation method by using mainly smartphone sensors. For dataset, we collect ECG (for label), 3-axis acceleration, pressure, illuminance, GPS and temperature (for training data) under different exercise types (walking and running) by using a smartphone and smart clothing called hitoe. To construct a regression model, we adopt a dual stage attention-based RNN model. From the evaluation results, we confirm that the proposed method can estimate RRI and LF/HF with acceptable accuracy.

Recently, an efficient and automatic infrastructure maintenance service is mandatory. To address this demand, in this paper, we introduce a segmentation-based road damage detection method by using U-Net. To train the model, we collect 4K images by using a smartphone mounted on a bicycle and build our own road damage dataset. In addition, to improve detection accuracy, we apply focal loss and image patch for loss function and input image, respectively. From the evaluation, the result confirms that the method demonstrates to extract road damages with acceptable accuracy.

In this paper, we propose a deep learning-based distance estimation method from a single-shot image. In the proposal, we model the estimation as a regression problem, and estimate the distance between a pedestrian and a camera by using three main features; size of bounding box, image blur and image features. By using KITTI dataset, we evaluate the accuracy of the proposed model.

Recently, in order to build barrier-free smart cities, Japanese government is actively promoting to construct open data regarding roads and buildings, including existence of bumps, road slope and corridor width. To contribute this activity, in this paper, we introduce a corridor width estimation method by using an RGB-D camera. In this method, we assume that the RGB-D camera is mounted on the electric wheelchairs for a mobile sensing platform. The corridor width estimation is mainly operated by three steps: planes extraction, side walls extraction and distance calculation. We evaluate the accuracy of this method at eight corridors in Waseda University.

To contribute to publishing pedestrian networks as open data, this paper introduces a curb height estimation method by using 2DLiDAR mounted on e-Bike. We develop a prototype system by using e-Bike, 2DLiDAR and other sensors, and collect actual curb data around Waseda University. From accuracy evaluations, the results confirm that the prototype can estimate road level differences and positions with high accuracy.

In video streaming on the internet, effective encoding recipes (i.e. bitrate-resolution pairs) are a main obstacle to deliver high-quality video streams. We developed a method to generate an encoding recipe that considers subjective visual quality with one just-noticeable difference (JND) distance. However, this method requires excessive computation time, which is not directly applicable for live streaming. In this paper, in order to provide a light-weight method for live streaming, we developed three acceleration techniques: resolution extrapolation, VMAF skipping and sampled objective measure calculation. These techniques are heuristic, but greatly contribute to reducing computational cost. Experimental results demonstrate that the proposed method achieves a significant reduction in computation time without significant effects on rate-JND characteristics.

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.

Named Data Networking (NDN) has been recently introduced as a new network architecture. The implementation of adaptive bit-rate video streaming on NDN is one of the most challenging tasks. The in-network caching feature of NDN could mislead a consumer to overestimate the end-to-end throughput for the next video segment. In this paper, we propose a strategy for more efficient rate adaptation by allowing Interest message adaptation in a hop-by-hop fashion. Moreover, the proactive content caching is designed to be utilized on the related router when a bottleneck on the transmission path was detected. The experiments using NDN-JS and DASH-JS demonstrate that our proposed solution provides better average stream bit-rate than the traditional rate adaptation method. Also, the link bandwidth consumption for video data transmission using the proposed scheme could be reduced efficiently.

Recently, to provide a low-latency mobile computing platform, Mobile Edge Computing (MEC) is proposed. In this paper, we first summarized the feature capabilities of MEC, such as content distribution and caching, computational offloading and multimedia Internet of Things (IoT). Then, to understand recent research efforts of multimedia MEC, we briefly highlight the research efforts in terms of above three capabilities: how to achieve edge caching in video distribution, how to schedule computational offloading to the cloud and how the communication quality degradation affects to the user experience of multimedia IoT. Finally, we addressed the emerging research issues of multimedia MEC to improve reliability and robustness of multimedia MEC.

Disaster detection is an important research topic and draws great attentions from both industry and academia. In this paper, we study the hybrid landslide detection system, which utilizes the video surveillance camera and multiple kinds of sensors and can detect the landslide automatically. Edge processing is adopted in this hybrid system to fuse the sensor data based on the Dempster-Shafer (D-S) theory, i.e. utilizing multiple sensors' information to calculate the possibility of the landslide. Edge processing can make faster decisions than the control center, the results of the edge processing are then used to schedule the sensor's transmission frequency and video transmission under the network constraints in this system. The simulation results show that the proposed scheme outperforms the competing schemes in typical network scenarios.

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.

In order to analyze system latency of edge computing for multimedia data processing, we propose a model of system latency of two different distributed processing scenarios: pipeline and parallel processing. We validate the model by using actual cloud services and human detection processing.

In this study, we propose a broadcasting and communication integrated platform that can retransmit 8K broadcasting contents from a TV receiving set to mobile devices. We then evaluate the QoS characteristics and the qualities of 8K contents in this platform. Evaluation results conclude that 8K broadcasting contents can be retransmitted to mobile devices smoothly and their qualities are nearly as high as those of assumed 8K broadcasting service in Japan.

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.

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.

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 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.

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.

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.

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.

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.

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.

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.

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.

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.