Investigating The Use Of Machine Learning in Conventional Heterodyned-Based Receivers

This paper explores the incorporation of machine learning algorithm in conventional heterodyned-based detection approach, with a goal of improving detection performance of conventional heterodyned-based receivers in the presence of signal-to-signal beating interference. The proposed machine-learning-based heterodyned detection approach is evaluated through software simulation, and the results obtained at this stage suggest that the inclusion of machine learning could significantly improve the detection performance of heterodyned-based detection approach when the minimum frequency spacing required between the reference tone and the wireless signal is not met, using relatively simple neural network architecture.