A Novel Approach to Multiagent based Scheduling for Multicore Architecture

Abstract

In a Multicore architecture, each
package consists of large number of processors. This
increase in processor core brings new evolution in
parallel computing. Besides enormous performance
enhancement, this multicore package injects lot of
challenges and opportunities on the operating system
scheduling point of view. We know that multiagent
system is concerned with the development and
analysis of optimization problems. The main objective
of multiagent system is to invent some methodologies
that make the developer to build complex systems that
can be used to solve sophisticated problems. This is
difficult for an individual agent to solve. In this paper
we combine the AMAS theory of multiagent system
with the scheduler of operating system to develop a
new process scheduling algorithm for multicore
architecture. This multiagent based scheduling
algorithm promises in minimizing the average waiting
time of the processes in the centralized queue and also
reduces the task of the scheduler. We actually
modified and simulated the linux 2.6.11 kernel process
scheduler to incorporate the multiagent system
concept. The comparison is made for different number
of cores with multiple combinations of process and the
results are shown for average waiting time Vs number
of cores in the centralized queue.