Machine Learning Aided Channel Estimation for Ambient Backscatter Communication Systems

Ambient backscatter has been a hot research topic ever since its birth in 2013. One open problem for ambient backscatter communication systems is individual channel estimation, which is the focus of this paper. In this paper, we design a communication protocol for the reader and the tag so as to obtain all the channel parameters. Specifically, we utilize expectation maximization (EM) algorithm, one typical machine learning approach, to design a semi-blind estimator so as to acquire combined channel parameters. We also obtain the uplink channel between the reader and the tag with a maximum likelihood (ML) estimator and estimate the downlink channel with superimposed pilots from the reader. In addition, we derive all the Cramer-Rao bounds (CRB) of the proposed channel estimators. Finally, simulation results are provided to corroborate our theoretical studies.