Characteristics of sequential detection in Cognitive Radio Networks

Sequential Analysis is an effective detection procedure for spectrum sensing in Cognitive Radio (CR) Networks. On average, given P_fa (probability of false alarm) and P_md (probability of mis detection) and low SNR regime, it requires less independent samples for the primary users (PU) detection comparing to the Neyman-Pearson (NP) test. This paper deal with the evaluation of the cumulants of the distribution (PDF) of random time of the sequential analysis, however, for simplicity we consider only Gaussian approximation. It is assumed that the PU and secondary users (SU) are sharing the same frequency bandwidth and for spectrum sensing SU´s apply incoherent diversity combining of diversity branches with fading described by the Generalized Gaussian (GG) model.