Learning-based algorithm for energy-efficient channel decision in cognitive radio-based wireless sensor networks

Cognitive radio-based wireless sensor network is the new paradigm in sensor network technology. It is a combination of the traditional sensor network and cognitive radio technology. Apparent challenge to this new sensor network outlook is the problem of energy efficiency. In this paper, we present the energy-efficient channel decision using reinforcement learning-based algorithm. The proposed algorithm is a learning-based algorithm in which a learning agent decides its action in a particular state based on its learned experience in the past. Hence, future decisions are based on reward or punishment obtained from previous actions. Results of simulations carried out shows that the proposed algorithm performs nearly 70% better in terms of energy-efficiency compared with random channel selection scheme.