Machine Learning-Assisted Detection for BPSK-Modulated Ambient Backscatter Communication Systems

Abstract

Ambient backscatter communication (AmBC), a green communication technology, is hampered by the continuously and extremely fast varying, strong and unknown ambient radio frequency (RF) signals. This paper presents a machine learning-assisted method for extracting the information of the AmBC device. The information is modulated on top of the unknown Gaussian-distributed ambient RF signals. The proposed approach can decode the binary phase shift keying backscatter signals encoded using Hadamard codes. This method extracts the learnable features for the tag signal by first eliminating the direct path signal and then correlating the residual signal with the coarse estimate of ambient signal. Thereafter, the tag signals are recovered by using the k-nearest neighbors classification algorithm. The recovered signals are decoded by a Hadamard decoder to retrieve the original information bits. We validate the performance using simulations to corroborate the proposed approach.