Distributed load balancing in a multiple server system by shift-invariant protocol sequences

Abstract

Ideally, many application systems for distributed users should be designed without requiring a centralized controller, for example cloud computing or wireless sensor networks. A fundamental challenge to developing distributed algorithms for these systems is load balancing, which is the focus of study in this paper. A common feature of these distributed algorithms is that routing decisions should be derivable without requiring much information from the system, probabilistic routing is one example coming to mind. In this paper, we propose a new routing strategy based on the idea of shift-invariant protocol sequences. We study this load balancing approach in the context of a queuing model of multi-server system. Our model and strategy can be applied to many practical systems, including wireless networks. Numerical studies were carried out to compare our strategy with other routing strategies such as probabilistic routing and random sequences routing. The results show that the proposed algorithm has better performance than these strategies.