Katsuichi Nakamura

Development of Arkas Sensor Platform and Application for Heavy Snowfall Observation

A sensor platform, which can accommodate various types of sensors and can extract the meaningful information from the sensor data, is necessary for building a sensor network system. In this study, we propose a new sensor platform, Arkas, which can be easily utilized for various environmental observations. The Arkas sensor platform consists of a sensor node which is designed for easy-to-use management provided by the Arkas server application over Android. In order to evaluate the practicality of the proposed sensor platform, a sensor network system for observing snow in Niigata-prefecture, Japan has been developed as an application of the platform. The system utilizes an ultrasonic distance sensor to derive an amount of snow cover and a snowfall from distance data. From experimental evaluations conducted from Dec. 2010 to Mar. 2011, we have confirmed that the system can accurately estimate the amount of snow cover and the snowfall by processing (e.g., noise filtering) the sensor data. As a result, we have concluded that the proposed sensor platform can be used for building a snow observation system and can support human-activities in a area of high snowfall.

Sensor network for observation of seabirds in Awashima island

It becomes so important to observe a wild life for obtaining not only knowledge of its biological behaviors but also interactions with human beings in terms of geoenvironmental investigation and assessment. A sensor network is considered to be a suitable and powerful tool to monitor and observe a wild life in fields. In order to monitor/observe seabirds, a sensor network is deployed in Awashima island, Japan. A sensor platform is useful for early and quick deployment in fields. Atlas, a server — client type sensor platform, is used with several sensors, i.e., infrared sensors, thermometers within a nest, a sound sensor and network camera. Although the project has just started, useful data such as a leaving/returning pattern of seabirds are figured out by the sensor network.

On demand routing scheme for real-spatial information based group communication

Widespread high-performance mobile stations and wireless communication technologies make it possible to enjoy various network services anywhere at any time. In particular, social network services (SNSs) provide a wide variety of communication among users. We aim to achieve real-space communication that directly integrates information exchange and real-space activity in our daily lives. Therefore, we develop three subsystems: (i) gathering personal and location information of users, (ii) managing huge quantities of user information and creating a group of users on a request basis, and (iii) delivering a message for all users in a requested group. In this paper, we focus on the third subsystem, especially in a combinational configuration of wide-area wireless networks (WiMAX, LTE, and so on) and device-to-device (ad hoc) communication with WLANs. We assume that some users who should receive a message cannot use wide-area wireless networks due their inaccessible communication range or the occurrence of a disaster, and that supplemental device-to-device WLAN communication covers such users. Although multicast communication can provide effective message delivery to users in a group, the overhead increases with the increased number of messages multicast to many users and groups. We propose an on-demand routing scheme for group communication based on a real-spatial information system and evaluate the efficiency of the proposed scheme by a simulation.

Field tests and indoor emulation of Real-Spatial Information Based Group Communication

We proposed a novel scheme that combines information exchange and user social activities in real-time so that users can integrate their communications seamlessly with their daily lives. Called the Real-Spatial Information-Based Group Communication (r-Space) system, this scheme creates r-Space groups by considering personal and location information and enables users to share information within those groups. We developed a prototype system consisted of client application and control server and then conducted a field experiment in Nov. 2016. In PerCom 2017, the introduction of the experiment and its live demonstration will be exhibited.

Behavioral Anomaly Detection System on Network Application Traffic from Many Sensors

For a computer network in the era of big data, we discuss a behavioral anomaly detection system which makes it possible to analyze and immediately detect anomaly traffic behavior. Many sensor devices connect to the network and tend to generate their application traffic at quite a low communication rate. In order to observe necessary traffic information for traffic analysis in a short time, the monitoring system integrates traffic statistics of flows sent from devices which are considered to generate the same application. It detects anomaly traffic behavior on the basis of application analysis using NMF(Non-Negative Matrix Factorization).

On the Framework for Network Measurement as a Service -- The perfSONAR-based Integrated Network Management

An integration of multiple-location, diverse-type, and long-term measurements has been considered a key means for continuous acquisition of the global status and for timely and cost-efficient identification of the causes of communication performance degradation. In addition, the measured data and analyzed results should be shared and reused for efficiency. Therefore, a global network measurement platform as a service is indispensible, which can provide integrated network monitoring and analysis functionality on-demand and can adapt to the purposes of the individual users (applications) and operators. In this report, we briefly introduce the design and implementation of an integrated network measurement system, together with a preliminary experiment on the Internet.

An NMF-Based Traffic Classification Approach towards Anomaly Detection for Massive Sensors

For a computer network in the era of big data, we discuss a behavioral anomaly detection system which makes it possible to analyze and immediately detect anomaly traffic behavior. Many sensor devices connect to the network and tend to generate their application traffic at quite a low communication rate. In order to observe necessary traffic information for traffic analysis in a short time, the monitoring system integrates traffic statistics of flows sent from devices which are considered to generate traffic for the same application. It detects anomaly traffic behavior on the basis of application analysis using NMF(Non-Negative Matrix Factorization). This paper describes a basic design of our prototype development.

Effectiveness of priority and path control schemes for grid applications

In grid computing environments, a resource control that considers not only the ability of computer resources but also available network resources is needed in order to execute parallel and distributed processing effectively. In particular, the execution performance of a grid application will sometimes decrease due to the congestion caused by multiple flows running on the same network path. Therefore, a network control scheme in grid computing environments should consider the multiple flows generated from its own application as well as from other applications. In this paper, we propose and evaluate its effect of a priority control scheme for multiple flows that are generated simultaneously, in order to reduce the execution time of the grid applications without reserving network resources. Moreover, a method for improving the effect of the priority control by using path selection is discussed, which distinguishes the priority flows from the other flows.

Performance Evaluation of Priority and Path Control Schemes for Grid Applications

In a grid computing environment, a grid resource control that considers not only the ability of the computer resources but also the available network resources is needed in order to execute parallel processing effectively. In particular, there is a possibility that the execution performance of a grid application decreases because of the congestion created by two or more flows that are generated on the same network path. Therefore, a network control scheme in a grid computing environment should consider the multiple flows generated from its own application, as well as from other applications. In this paper, we propose a priority control scheme of several flows that are generated simulteneously, in order to improve the execution time of the grid application without reserving a network resources. Moreover, we discuss a method for improving the effect of the propriety control by using path selection that is distinct from the priority flows and other flows.