Improved energy detectors with data/decision fusion of partitioned time-domain samples

This article proposes several new and improved energy detectors for detecting unknown deterministic signals perturbed by zero-mean Gaussian noise and/or fading. In contrast to the conventional total energy detector (that computes the signal energy by summing the square of all time-domain signals), we first partition the time-domain samples in pairs and then combine the signal energy estimates from each of these partitions using different rules (i.e., data fusion) to develop new detectors that provide varying levels of primary user (PU) signal protection. Both numerical and simulation results show that all of our modified energy detectors outperform the conventional total energy detector. Decision fusion of the local binary hard decisions made at the sample partitions is also considered to provide a greater flexibility for the PU signal protection and/or to improve the detection performance of the stand-alone energy detector. The efficacy of cooperative spectrum sensing using the average energy detector is also studied.