Takayuki Shizuno

Statistical traffic identification method based on flow-level behavior for fair VoIP service

Many threats, such as denial-of-services (DoS) attacks and impersonation, can endanger an IP network where real-time communication, as represented by VoIP, requires a high level of reliability to work as a replacement for PSTN. Without making a concrete framework to protect the reliability of VoIP, we may soon start seeing cases whereby networks are illegally used by impersonating real-time communication. This paper proposes a technique, based on flow-level behavior, which aims to maintain communication confidentiality and identify applications of real-time communication, in a short period of time in order to eliminate applications that illegally use network resources. The paper shows favorable identification results were obtained from our experiments and also introduces issues we have discovered that must be resolved.

Wireless LAN access network system for moving vehicles

Several Wireless LAN handover schemes have been proposed for access networks. These efforts have not yet, however, produced a satisfactory solution for access networks that provide broadband Internet services to moving vehicles such as express buses, trucks and high-speed trains. In this paper, we propose a WLAN access network system for high-speed moving vehicles. The proposed system achieves a short handover time and no packet loss during handover. A daisy-chain network architecture reduces construction costs to enable the wide deployment of many WLAN access points. We also present a performance evaluation of the prototype system.

Development of Battery and Charger Integration System (BCIS) for EV charging stations

The popularization of electric vehicles (EVs) will give rise to challenges that include expenses for capacity upgrades, power supply shortages, and long waiting time to charge. We have therefore developed a Battery & Charger Integration System (BCIS) to support the burgeoning infrastructure for EVs. The system enables charging time reductions in simultaneously multiple EVs charging by shifting the charging power peak and using power of stationary batteries. This paper outlines the BCIS and the results of an experiment and simulation of charging time reductions.

A Lightweight Application-Layer Signaling Protocol

The session Initiation Protocol (SIP) is a widely used application-layer control protocol. The processing load of the SIP protocol is reportedly seen as one of the potential problems which may limit the range of its application although it has a number of advantages. Many research efforts have been made and have come up with various compression algorithms to reduce the packet size in order not to waste scarce bandwidth especially for wireless access networks. However, they increase the processing load at the same time. We propose a lightweight application-layer signaling protocol (lightweight-SIP) which reduces network bandwidth without sacrificing CPU performance by introducing a gateway which translates SIP into the proposed protocol. The gateway works as a cache for static and lengthy header fields of SIP messages. We expect the proposed protocol to be used by low-spec devices such as sensor nodes, less-expensive surveillance cameras and wearable computers so that it could reduce the energy consumption and lengthening the battery life or loosen requirements for the hardware specs.

Comparison of Data-Searching Algorithms for a Real-Time Information-Delivery System

The number of content-generation devices is expected to increase explosively in the near future thanks to evolution in the field of input/output devices and Consumer Generated Media (CGM) tools. Users will simultaneously wear communicating sensor devices wherever they are, whenever they need them,and whatever they are doing. Under such circumstances, various contents will be generated,updated by the devices and moved much more frequently than they are now. We propose an architecture for a real-time information-delivery system that enables users to quickly discover the freshest, most accurate content. We compared three types of well-known data searching algorithms, i.e., in DNS-like static-tree method, Web crawler method,and Chord method, which can be used in servers in existing systems to retrieve the metadata on the content.We first developed a simulator and evaluated the algorithms to assess the methods to find which generation, update, and movement of content occurred more frequently. We discovered that the Web crawler method and Chord method both offered roughly the same correct answer rate. However, the number of messages increased dramatically with the Web crawler method when the content-update interval shortened.The average search time also increased with the Chord method as the number of user nodes increased.