2003 Fiscal Year Final Research Report Summary
Distributed Algorithms and Applications for Allocating Resources Competed for by Process Groups
| Project/Area Number |
14550365
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
情報通信工学
|
|---|
| Research Institution | The University of Aizu |
|---|
Principal Investigator |
ZIXUE Cheng The university of Aizu, The School of Computer Science and Engineering, コンピュータ理工学部, 教授 (40254049)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
TONGJUN Huang The University of Aizu, Information systems and Technology Center, 情報処理センター, 助手 (30254121)
KOYAMA Akio Yamagata University, Faculty of Engineering, 工学部, 助教授 (60315679)
|
|---|
| Project Period (FY) |
2002 – 2003
|
| Keywords | Distributed Algorithms / Resource Allocation Deadlock / Starvation / Squeezing Technique / Concession Policy / Distributed Applications / Channel Allocation / チャンネル割り当て |
|---|
| Research Abstract |
Distributed Resource Allocation is an important for construction of distributed systems. There are a lot of researches on this problem. With the progress of computer networks, a variety of distributed cooperative applications are necessary. For distributed cooperative applications, cooperative processes can be considered as a group. When more than one group perform their works respectively in a network, processes belonging to different groups may compete for a resource. Careless allocation of resources may lead to deadlock between groups or starvation of a group. In order to solve such problem, we have define a newproblem, resource allocation to process group, and given several distributed algorithms for solving the problem. In this Project, based on research results in the past, we first proposed an efficient allocation method using unobstructed squeezing technique, such that the squeezed processes will not be affected i.e.to wait longer than without squeezing. The total efficiency of using the resources can be increased by the technique. Then we studied how to reduce the average waiting time of processes which request a resource by using a concession policy. That is, a process may cancel a reserved resource and rereserve another one, in order for another process to use the former. Moreover, we propose a new general model for resource allocation among groups which may execute a task concurrently. Finally, we discussed if those algorithms can be used for some applications, and considered how to implement an algorithm for allocating channels of mobile phones.
As a future task, we plan to incorporate the developed squeezing method and concession method into the model for resource allocation among groups.
|
Research Products
(4 results)
