2005 Fiscal Year Final Research Report Summary
On the Stability of Huge-scale Distributed Systems - the Advent of the Era of Tera
| Project/Area Number |
14380145
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
計算機科学
|
|---|
| Research Institution | KYUSHU UNIVERSITY |
|---|
Principal Investigator |
YAMASHITA Masafumi Kyushu University, School of Information Science and Electrical Engineering, Professor, 大学院・システム情報科学研究院, 教授 (00135419)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
SADAKANE Kunihiko Kyushu University, School of Information Science and Electrical Engineering, Associate Professor, 大学院・システム情報科学研究院, 助教授 (20323090)
ASAHIRO Yuichi Kyushu Sangyo University, Faculty of Information Science, Associate Professor, 情報科学部, 助教授 (40304761)
ONO Hirotaka Kyushu University, School of Information Science and Electrical Engineering, Research Associate, 大学院・システム情報科学研究院, 助手 (00346826)
|
|---|
| Project Period (FY) |
2002 – 2005
|
| Keywords | Huge distributed systems / stability / probabilistic methods / global information / local information / distributed algorithms |
|---|
| Research Abstract |
Unlike the conventional concept of distributed system, by a distributed system, we mean a one consisting of autnomonous agents that can communicate with each other by some communication devices, including for example a system of autonomous mobile robots, not only usual distributed systems on computer networks. Some huge distributed system like WWW consists of several GIGAs of agents and the era that the size of a distributed system is counted by TERAs is just around the corner.
Targets of this research are such huge distributed systems. The huge distributed systems can be characterized as follows ; first, the topology changes so rapid that each agent cannot ignore it, but second, the amount of local information that each process can collect relatively shrinks compared with the amount of global information.
The objectives of the research are to establish the concepts of the stability of the system and the correctness of the protocol, and propose methods to design correct protocols to keep the stability of a given distributed system.
In the following, we explain results of this research project by partitioning them into three groups. The first group contains papers on probabilistic search methods for huge distributed systems. A highlight among the results is about random walks using local information ; those random walks achieve the optimal hitting time and hence this local information is as powerful as the global information, as long as the hitting time is concerned.
The second group consists of papers on stabilizing protocols for huge distributed systems. We in one of the papers introduced the concept of stabilization for dynamic distributed systems, and presented a stabilizing token circulation protocol. The third collects search problems on plane, which appears frequently in robot systems. Note that papers in the publication list are selected from the works done in 2005.
|
