Channel estimation for opportunistic spectrum sensing: Uniform and random sensing

The knowledge of channel statistics, as a result of random fading, interference, and primary user activities, can be very helpful for a secondary user in making sound opportunistic spectrum access decisions in a cognitive radio network. It is therefore desirable to be able to efficiently and accurately estimate channel statistics, even for resource constrained secondary users like wireless sensors. In this paper we focus on the traditional ML (maximum likelihood) estimator. However, rather than using equal or uniform sampling/sensing intervals as is typically done, we introduce a random sampling/sensing based ML estimation strategy. The randomization of the sampling intervals allows us to catch channel variations on a finer (time) granularity; the associated likelihood function is also more sensitive to channel variations. Consequently, this scheme significantly reduces the average sampling rate compared to uniform sampling. Analysis and simulation both show that random sampling significantly outperforms uniform sampling at low sampling rate. We further propose a randomized uniform sampling scheme which achieves a better tradeoff between good performance of random sampling and the low complexity of uniform sampling.