Automatic TM-CFAR Based on Artificial Neural Networks: Application on the Acquisition Stage of DS-CDMA Systems.

Abstract

The primary objective of this study is to master and assess the performance of the Adaptive Trimmed mean Mechanism Constant False Alarm Rate (ATM-CFAR) detector. The system under investigation utilizes a straightforward serial search strategy and incorporates an adaptive detector based on multilayer artificial neural networks (ANNs), trained via the backpropagation algorithm. ANNs are computational models inspired by the human brain, capable of recognizing patterns and making decisions. These networks consist of interconnected layers of nodes (neurons) that process input data, learn from it, and improve detection capabilities over time. Our findings underscore the critical role and efficiency of systems employing automatic censoring algorithms in the acquisition process. Moreover, the proposed system demonstrates a superior detection probability, affirming its suitability for advanced applications across diverse technological fields. The study comprehensively explores the integration of neural networks in adaptive detection, highlighting significant improvements in detection capabilities and overall system performance.