ABSTRACT
Ambient Backscatter Communication System (ABCS) leverages ambient Radio Frequency (RF) signals (such as television, FM radios, Bluetooth, cellular or wireless fidelity) to backscatter the received RF signal with the information bits from a battery-free device(a tag or a sensor) to the reader device, allowing two passive devices to exchange their information with each other. ABCS optimizes real-time monitoring, extends data collecting stand-by time, and reduces size, cost, and power consumption in the Fifth Generation (5G) Internet of Things (IoT) network. Recently, Machine Learning (ML)-based techniques for analysis of ABCS have attracted huge attention from both industry as well as academia. In particular, the performance analysis of ABCS using supervised, unsupervised, and reinforcement learning is found in the literature. In unsupervised ML, K-Nearest neighbor classification technique is used to identify the received information bits by clustering them based on the received energy characteristics at the reader end. Reinforcement ML is being applied to improve the performance of the backscatter system. The learning aspect of the system is being developed, resulting in a battery-free and intelligent sensor device for the IoT network. Finally, in this chapter we outline the research trends in ML-based ABCS.
