2008 Index IEEE Transactions on Systems, Man, and Cybernetics - Part A: Systems and Humans Vol. 38

Abstract

Wireless sensor networks are characterized by energy-constrained nodes that are tasked with collecting and forwarding environmental parameters with a requisite measurement, which is spatial and temporal fidelity. At the system level, fidelity is not the only issue of interest but also the achievement of a low-cost solution and a long life for the deployed network. As such, sensor nodes should be low in complexity and should achieve the requisite fidelity requirements, with minimum communication and coordination. This paper proposes that these nodes can operate as automata and still achieve the overall system performance requirements with minimal control. This paper presents and analyzes an automaton architecture and a control strategy designed to maintain spatial fidelity as the performance objective. In particular, we show the following: 1) that the architecture permits control of the number of nodes actively transmitting information in each epoch (denoted by Q); 2) that the variance of Q can be controlled and, particularly, can be set to a value significantly less than that of a Bernoulli-process benchmark (i.e., the architecture is expedient with respect to the control of this variance); 3) that the control strategy is scalable over several orders of magnitudes; and 4) that the methodology is efficient in approaching benchmark performance with respect to energy usage. The proposed methodology has the following specific advantages over the benchmark: 1) The total number of sensors deployed in the network need not be known, and 2) the strategy maintains a robust control of Q over changes in the commanded value and changes in the number of deployed sensors.