Energy-aware channel selection for cognitive wireless sensor networks

Wireless sensor nodes operating in unlicensed bands can exploit frequency agility in order to mitigate, through dynamic spectrum access, the effects of interference induced by collocated networks. With this respect sensor devices can select the frequency band used for their communications in an opportunistic manner: this however requires spectrum sensing and consumes precious energy when more and more channels are sensed. In this paper we propose an algorithm allowing to limit the overhead introduced by spectrum sensing. We formulate a stochastic control problem which solution allows sensor nodes to decide whether it is better to stop the sensing procedure and transmit on a sensed channel or check a new frequency band hoping to find better interfering conditions. Simulation and experimental results show significant energy gains with respect to previously proposed channel selection algorithms that sense all the available channels.