Real-time applications, such as ticket booking systems, video streaming systems, social networking services, and online games, are prevalent nowadays. Low latency between a user and a server is required in such applications. A distributed server approach is used instead of a single server one to achieve this. In such a situation, selecting which users connect to which servers significantly determines the system’s performance. There is a problem with the distributed server approach; each server may fail due to maintenance or breakdown. Single-server failure is more popular than multiple-server failure, so this paper focuses on single-server failure. In this paper, we assume that the capacity of each server is variable and minimize the total capacity of the servers. We also set the assumption that each user is allowed to connect exactly two servers. Under this assumption, we select each user’s main and backup servers. This paper proposes two approximation algorithms for a server allocation problem to minimize total capacity under server failure. We formulate the problem as an integer linear programming (ILP) problem. We show that the proposed algorithms have an approximation rate of$4/3$. One of them is an improved version of the other one. The algorithms are based on a 2-approximation algorithm of a vertex cover problem, in which taking a maximal matching is the key idea. Numerical results show that the proposed algorithms run significantly faster than ILP. The performance of the proposed algorithms depends on the topology; still, in randomly generated general networks, it increases the total capacity by 2.03% or less than ILP on average.

MOTIVATION: Single-cell RNA sequencing (scRNA-seq) analysis reveals heterogeneity and dynamic cell transitions. However, conventional gene-based analyses require intensive manual curation to interpret biological implications of computational results. Hence, a theory for efficiently annotating individual cells remains warranted. RESULTS: We present ASURAT, a computational tool for simultaneously performing unsupervised clustering and functional annotation of disease, cell type, biological process and signaling pathway activity for single-cell transcriptomic data, using a correlation graph decomposition for genes in database-derived functional terms. We validated the usability and clustering performance of ASURAT using scRNA-seq datasets for human peripheral blood mononuclear cells, which required fewer manual curations than existing methods. Moreover, we applied ASURAT to scRNA-seq and spatial transcriptome datasets for human small cell lung cancer and pancreatic ductal adenocarcinoma, respectively, identifying previously overlooked subpopulations and differentially expressed genes. ASURAT is a powerful tool for dissecting cell subpopulations and improving biological interpretability of complex and noisy transcriptomic data. AVAILABILITY AND IMPLEMENTATION: ASURAT is published on Bioconductor (https://doi.org/10.18129/B9.bioc.ASURAT). The codes for analyzing data in this article are available at Github (https://github.com/keita-iida/ASURATBI) and figshare (https://doi.org/10.6084/m9.figshare.19200254.v4). SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

When a natural disaster occurred, some roads cannot be used anymore and sometimes blocked. Also, survivors and refugees cannot follow the evacuation procedures by just using default maps after disaster. A previous study proposed a safety route guidance system that can be used after natural disasters by using participatory sensing. The system estimates safe routes and generates an evacuation map by collecting GPS data and accelerometer data from pedestrians’ smartphone. However, the system does not base on default map data. After that, the system evaluates the safety of each route. However, the previous study did not propose a method of finding evacuation routes from the users’ current location to their destination. Therefore, in this study, we proposed a method of evacuation route planning. We have implemented Multi-Objective Genetic Algorithm (MOGA) into the route planning methodology. The proposed system has three objective functions, which are: evacuation distance, evacuation time and safety of evacuation route. Also, we proposed a new safety evaluation method. As a result, this study gives a better reflection of the change of road conditions. Also, the safety evaluation values are more useful than the previous study's evaluation method of the route. Moreover, the system can provide evacuation routes with different characteristics to users. As a result, the users can select a route which is suitable for their situation.

The number of elderly people living alone has been increasing in Japan. If dementia has been detected at an early stage, the progress of disease can be slowed. Therefore, early detection of dementia is so beneficial to the patient therapy. Most of dementia symptoms can be detected by housemates, noticing some changes of behavior in elderly people. However, it is difficult to detect early symptoms of dementia in elderly people living alone. In this study, we proposed a system which detects the symptoms of the elderly people living alone by using the M2M (Machine-to-Machine) /IoT (Internet of Things) platform. We installed sensors inside the house of elderly people living alone. These sensors can detect behavior by initial symptoms of the dementia. Also, these sensors’ data will be used as behavioral data of dementia patients. In addition, we conducted a questionnaire about their habits and personalities. These questionnaires’ results were used as an attribute data of the subjects. The system analyzes the two data and then determines the presence or the absence of dementia.

WIDESTAR II is a mobile satellite communications service that covers the landmass of Japan and the surrounding maritime area using the N-STAR c/d geostationary satellites, which are positioned approximately 36,000 km above the equator. The system is designed to provide highly reliable communications during disasters and diverse satellite communication services over the coming ten or more years. To satisfy user demand, we developed a mobile satellite access scheme that provides a higher transmission rate and that flexibly accommodates the latest IP technology. The access scheme is based on accumulated experience based on mobile satellite communications technology combining the 3G cellular-based scheme. We also developed base station equipment that is connected to the core network equipment based on IP multimedia subsystem (IMS) to achieve an all-IP system.

The tactile sensor is required for various robots. In humanoid robots, flexibility of the sensor is an important feature for preventing physical damage and for interacting with the human. Moreover, stretchability of the sensor has advantages that the sensor is nonbreakable and that the sensor can be easily mounted on curved surfaces or deformable parts such as joints. This paper proposes a novel tactile sensor made of flexible and stretchable silicone rubber. A structure of the sensor is similar to the capacitive tactile sensors. However, the proposed sensor utilizes a different principle from existing sensors. The sensor utilizes static electricity and electrostatic induction phenomenon, and can detect some touch conditions. This paper reports the principle and characteristics of the proposed sensor. Experiments show that the sensor output depends on touch area, touch velocity, and material of touch objects. However, the sensor does not depend on touch weight. Moreover, the experiment shows that even if the proposed sensor is stretched, it performs as the tactile sensor.

Mobility support for mobile networks will be more important as the mobile Internet becomes increasingly popular. To support mobile networks, the concept of prefix scope binding (PSB) is being discussed in IETF; however, by only applying this concept to Mobile IPv6 (MIP), the problem of packet loss still remains. In this paper, we apply the PSB concept to our proposed protocol, the hierarchical Mobile IPv6 extension with buffering function (HMIP-B), in which mobility anchor points buffer packets destined to the mobile hosts during handoff. We compare MIP and HMIP-B based on the handoff performance related to mobile networks by computer simulation. The simulation results indicate that our proposal improves the handoff performance in both TCP and UDP communications.

The paper proposes an autonomous decentralized building control system to cope with the requirements of building control systems under multi-vendor environments. In the proposed system building equipment is horizontally distributed on a network and autonomously works by sharing communication data on the network according to objects for building control systems. The proposed system deals with the requirements of building control systems and offers its users, such as building designers, a user-friendly tool for the system construction. The authors call such a communication platform an autonomous decentralized system (ADS) platform. By utilizing the ADS platform, the building control system obtains two ADS properties, i.e., autonomous controllability and autonomous coordinability. Furthermore the proposed system can realize the on-line expandability, on-line maintainability and fault tolerance of the system by relaying messages between the building system objects on the ADS platform without specifying the sending addresses.

We make a comparision of classification ability between BPN (BackPropagation Neural Network) and k-NN (k-Nearest Neighbor) classification methods. Voice data and patellar subluxation images are used. The result was that the average recognition rate of BPN was 9.2 percent higher than that of the k-NN classification method. Although k-NN classification is simple in theory, classification time was fairly long. Therefore, it seems that real time recognition is difficult. On the other hand, the BPN method has a long learning time but a very short recognition time. Especially if the number of dimensions of the samples is large, it can be said that BPN is better than k-NN in classification ability.

A system that reconstructs 3D models from hand-drawn three-view drawings represented by binary images is described. The authors lay main stress on the description of a heuristics directed reconstruction algorithm. The proposed algorithm performs a combinatorial search based on the face decision strategy along with two heuristics aiming at simulating the human's understanding way of interpretation. One is for finding a good sequence of search nodes, and the other for finding more natural scenes earlier than unusual ones.>