Resource Allocation and Emergent Coordination in Wireless Sensor Networks

Coordination in wireless sensor networks (WSN) is required for many tasks that are best achieved collectively, such as coverage and medium access. One of the major challenges in the design of WSN are the strong limitations imposed by finite onboard power capacity. Because communication requires considerable energy, it is imperative to have a coordination mechanism that requires little or no communication. Moreover, since WSN are likely to operate in unstructured and dynamic environments, the coordination mechanism has to be adaptive and robust with respect to environmental changes. Lack of centralized control in WSN requires alternative means for coordinating actions and resources of individual nodes to achieve long network lifetime, while not severely compromising network task performance. In this paper we explore the paradigm of emergent coordination as a mechanism for adaptive, distributed coordination in WSN. Specifically, we study a WSN composed of self–interested nodes that utilize a simple reinforcement learning scheme and achieve coordination by playing repeated resource allocation (load balancing) games with changing resource (load) capacities. Our results indicate that for a certain range of parameters the network is very adaptive to these changes. Although we formulate the problem in rather abstract settings of repeated games, the methods can be applied to a range of specific sensor coordination problems such as network coverage