Agent-Based Dynamic Scheduling Using Bidirectional Announcement Mechanism in Distributed Systems

In this paper, we propose a novel agent-based scheduling strategy ABDS using bidirectional announcement mechanism (BAM) in which an improved contract net protocol (CNP) is considered. Both the forward announcement and backward announcement in BAM perform the collaborative process, and jointly accomplish the scheduling. Besides, the calculation rules of bidding values are investigated to solve the scheduling issue in our study. Moreover, the heuristics to select contractors are proposed. The ABDS strategy employs the fair competition principle of roulette wheel and dynamic adjustment principle of buffer pool to solve the problem of local searching and to improve the load balancing of resources. We conduct extensive experiments to validate the superiority of our ABDS by comparing it with a unidirectional announcement (UA) scheduling algorithm. Also, the sensitiveness of priority parameter to the system performance is evaluated. Experimental results show that ABDS significantly improves the scheduling quality of UA and it is suitable for real-time task scheduling in distributed systems.