Comparison of energy detection using averaging and maximum values detection for dynamic spectrum access

Opportunistic Dynamic Spectrum Access (DSA) systems are emerging as a key to enable the Department of Defense (DoD) to meet its technology requirements for access to the electromagnetic spectrum. One of the main goals of DSA is to protect incumbent spectrum users from interference that could be caused by DSA emitters. One of the ways to accomplish this is through spectrum sensing as means to enable the DSA system to detect the presence of emissions from other spectrum users. A simple and practical approach to spectrum sensing is energy sensing that detects other users on the basis of the total signal energy in channel of a given bandwidth. In this paper our goal is to investigate the performance of energy sensing for constant and random power signals using two approaches. The first approach that has been investigated extensively in the past uses the total energy, or equivalent average power, over the entire signal bandwidth as the basis for detecting incumbent signals. The second approach, which needs to be used for wideband signals if the bandwidth of the incumbent signal is not known, uses signal energy in individual FFT bins as the basis for detection. We examine the performance of these detection approach under the Neyman-Pearson criterion and also using the Chernoff-Stein Lemma to obtain the achievable detection probability as a function of detection delay. We find that the method using average over the entire signal achieves better performance than the method that uses values of individual bins. This shows the importance of prior knowledge that can be expressed as radio policy to enable coexistence of DSA with incumbent systems.