Low-overhead Heuristic Algorithms for Spectrum Sensing in Cognitive Radio Networks

Cognitive Radio (CR) has matured as a research field and has the goal to improve the utilization of spectrum. This technology enables a CR user to function in unlicensed or licensed bands without causing interference to primary users. In order to achieve this functionality a CR user needs to sense the spectrum and exploit efficiently the transmission opportunities. In this paper we propose two heuristic sensing algorithms for adhoc CR networks with low computational requirements, where each user tries to exploit ??spectrum holes?? based only on its own observations. The first algorithm refers to an immediate reward approach, according to which, each time a ??spectrum hole?? is detected, it is used instantaneously. The second refers to an immediate-versus-future-reward approach, where the CR may wait to sense the next channel instead of transmitting in the current one, if the expected reward when using the next channel is greater than the instant. In our proposed algorithms we examine both known and unknown statistics of channels. For the case of unknown channel parameters we introduce an initial sensing time and we investigate through simulations the impact of different sensing periods on the system's throughput.