Threshold Activation Policies in a Random Sensing Environment

Under Markovian assumptions, the performance of threshold activation policies are studied, for two different correlation models of the energy discharge and recharge processes, to show that the optimal threshold policy guarantees performance within factor of 34 of the optimum over all possible policies.

Abstract

We consider the problem of how sensor nodes should be activated in so as to maximize a strictly concave utility function in a random sensing environment. Sensors are assumed to be energy-constrained, but rechargeable, and the energy discharge and recharge times are random. Under Markovian assumptions, we study the performance of threshold activation policies, for two different correlation models of the energy discharge and recharge processes. For both models, we show that the optimal threshold policy guarantees performance within factor of 34 of the optimum over all possible policies. We also comment on the effect of the correlation model on the performance of the threshold policies.