A Simplified LLR-Based Detector for Signals in Class-A Noise

The design of a simplified detector for signal in Middleton´s class-A noise is considered. The optimal detector is impractical due to the complexity of the probability density function of the noise. The conventional Gaussian detector (known as the matched filter or the correlator) has near-optimal performance only with relatively high SNR values. Different suboptimal detectors have been proposed to give robust performance with different levels of complexity such as the locally optimal Bayesian detector. In this paper, we propose a unified simple approach to design a near- optimal detector with considerably low complexity by linearly approximating the optimal log-likelihood ratios of the received symbols. The resultant detector has near-optimal performance with low complexity.