A study on dynamic adaptation of distributed systems using immune mechanism
| Project/Area Number |
12680359
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
計算機科学
|
|---|
| Research Institution | Keio University |
|---|
Principal Investigator |
YAMAMOTO Yoshikazu Keio University, Faculty of Science and Technology, Associate Professor, 理工学部, 助教授 (20051873)
|
|---|
| Project Period (FY) |
2000 – 2001
|
| Project Status |
Completed (Fiscal Year 2001)
|
| Budget Amount *help |
¥3,200,000 (Direct Cost: ¥3,200,000)
Fiscal Year 2001: ¥1,300,000 (Direct Cost: ¥1,300,000)
Fiscal Year 2000: ¥1,900,000 (Direct Cost: ¥1,900,000)
|
|---|
| Keywords | dynamic adaptation / artificial immune network / meta-level architecture / Web server / network server / fine grained object / self-organized system / design pattern / Webサーバ / ネットワークサーバ / 自己反映計算 |
|---|
| Research Abstract |
It is very interesting problem that how to make a software used in dynamically changing environment or usage patterns, such as a server that is used by public or unspecified systems. If the dynamic environmental change is predictable, it can be achieved by incorporating these conditions and actions as production rules into the software. On the other hand, for the software such as Web servers where access patterns are completely unpredictable, it is highly appreciate for system developers if the system has an ability to adapt autonomously to dynamic environmental change.
The purpose of this research was to implement a mechanism that is necessary for dynamic adaptation of systems. We utilize fine grained objects, reflection and meta-level architecture as base technologies. In the first year, we developed a policy coordination engine named iNexus that utilized an artificial immune network as a tool for autonomous and dynamic adaptation to the environment. And then we implemented OpenWebServer that utilized the iNexus, and verified that the OpenWebServer could change processing policies depending on dynamically changing access patterns. In the second year, we noticed high adaptability to environmental change of an artificial immune networks and developed a communication management system that used an artificial immune network as a tool for decision making of control policy. The system also showed high applicability of the artificial immune network even to the real application.
In case of a distributed system as a whole dynamically adopts to its environment, we realize a mechanism that makes copy of contents of a server which has over concentrated access. This mechanism can be used for a aid of congestion avoidance tool of Web access in the Internet. We also study the cost of copy and communication load in various environments and verify its usefulness in practical application to Web servers.
|
Report
(3 results)
Research Products
(13 results)
