Akira Chugo

A policy based networking architecture for enterprise networks

We describe a new concept for policy based systems management (PSM) and propose its component functions. PSM aims to reduce the total cost of ownership (TCO) while providing highly functional enterprise systems that include enterprise networks. Conventional enterprise systems have been managed manually by humans, and are not able to provide a highly functional service with low TCO. For this reason, the policy based networking (PBN) approach is gaining attention. This paper describes the PSM that extends PBN for total systems management and shows the effectiveness of PSM functions with an example of dynamic QoS configuration.

Autonomous routing scheme for large scale network based on neural processing

The authors discuss an autonomous routing scheme (called Holonic Routing) for large-scale networks that uses the principle of neural networks. In this scheme each node computes the optimal route whenever it tries to transmit a packet; thus the network can always maintain a balanced load, providing maximum transmission capacity in response to changes in traffic and topology. Because the routing procedure is packet by packet, even if a failure occurs in transit the packet will automatically be routed to avoid failed parts. Thus, a reliable nonstop network can be easily constructed using this scheme. The scheme is especially suitable for multimedia networks where large capacity and high reliability are needed.<<ETX>>

Voice quality management for IP networks based on automatic change detection of monitoring data

Recently, quality management of IP networks has become an important issue with the development of real-time applications such as IP-phones, TV-conferencing and video-streaming. Specifically, when voice data is mixed with various application data, there are worries that there will be a critical degradation in voice quality. In current management of voice quality, the general scheme is to monitor the voice quality using a preset threshold for the data. However, its not easy to set a proper threshold for data monitoring in each case, because the characteristics of the monitoring data become more diverse as networks become larger in scale and more complex. We propose an automatic change detection scheme that detects changes in the behavior of the monitoring data with changes in the traffic conditions, without any preset threshold. The essential concept of our scheme is the application of a statistical scheme to sequential data monitoring. We also demonstrate the effectiveness of our proposed scheme when applied to voice quality management.

Holonic routing scheme based on neural computations

An autonomous routing scheme based on neural processing is described. By using the properties of fast computation and self-organization of neural networks, the scheme can obtain an optimal packet direction well suited to traffic loads changeable with time. The neural network used in this scheme consists of neurons placed at each node in one-to-one correspondence to the input/output link. The performance evaluation using computer simulations shows that this scheme has good characteristics even in an unbalanced traffic environment, as compared to previous autonomous routing schemes.<<ETX>>

Development of terabit‐class super‐networking technologies

We propose terabit-class super-networking technologies, designed to improve the scalability, reliability and performance of optical Internet protocol networks. Our technologies comprise both intra- and interlayer traffic engineering technologies. The intralayer technologies include an optical path protection scheme, an electrical load-balancing scheme and a distributed content-caching scheme. These provide an effective and economical way of improving performance and reliability. The interlayer technologies include both traffic-driven and application-driven optical cut-through control schemes and a policy control scheme. These provide an effective and economical way of improving scalability and performance. The feasibility of our technologies has been verified by means of experiments using prototype systems. The results showed that the different techniques can be combined to form a single network architecture for dynamic optical path control. Copyright

A Scalable Resource Management Mechanism with Feedback Control for Network Systems

With the progress of large-scaled networks, a service platform should manage effectively enormous amounts of diversified information to ensure stable operations of the platform systems in a resource-constrained environment. We propose a resource control mechanism based on a feedback control law to enable effective resource management and the stable operation of platform systems. Furthermore, we introduce the application cases for a session admission control system and a packet sampling control system, and demonstrate the effectiveness of the proposed mechanism in simulations and prototype evaluations.

Distributed cache management for context-aware services in large-scale networks

In recent years, the number of messages transferred through networks has skyrocketed with the rising number of network nodes. For example, servers collect variant sensor information and store it. Although related works exist that present cache servers intended to reduce network costs, those cache functions do not work well for short lifetime content because most cached contents with short lifetimes are expired before referral. In this paper, we propose a method to resolve the problem for the coming ubiquitous network society, which is an asynchronous cache management method according to the application requirement. This method enables the CPU load of the servers to be reduced through service and application management for short-lifetime content, too. Furthermore, we propose a load-balancing method using autonomous message exchange instead of a management system. This method enables the CPU load to be balanced over multiple servers.

Dynamic routing for bursty traffic in ATM networks

The ATM-based broadband ISDN (B-ISDN) has been extensively studied recently . The purpose of ATM (Asynchronous Transfer Mode) is to support a variety of medi a efficiently, guaranteeing the quality of each service by unified cell-based switching . B-ISDN is expected to support a broad range of media such as constant-rate voice/video, variable-rat e voice/video, large-volume imaging and graphics and high-speed interactive data . The man y different requirements these media have in bandwidth, burstiness, and service quality, mak e unified traffic control very difficult in multimedia-integrated networks .We propose in this paper a traffic management approach that divides bandwidth to two classes -one for real-tim e connection-oriented traffic and one for non-real-time connectionless traffic -and manages eac h class using independent bandwidth control .