Research On Maximization Of Backscatter Communication Rate Based On Reinforcement Learning In Fading Channels

Radio frequency identification technology is the key technology to realize the Internet of Things.It is an automatic identification technology that can realize the contactless information transmission by using radio frequency signals to achieve object recognition.However,in recent years,as people's urgent need for the Internet of Things continues to escalate,there are still many limitations to the widespread use of RFID technology in the Internet of Things.For traditional passive tags,the communication distance is too short and needs to be A specialized reader to assist communication does not have sufficient portability;for active tags,the battery greatly increases the size and weight of the tag,and the manufacturing cost and the cost of periodically maintaining the battery are also relatively expensive.The environmental back reflection technology is a new technology derived from radio frequency identification technology,and it is also a new communication method that has emerged in recent years.In particular,ambient backscatter communication enables wireless devices to communicate without requiring active radio transmissions.It can utilize the radio frequency signals existing in the surrounding environment,such as Wi-Fi signals,broadcast signals,etc.,and backscatter tags use these signals as carriers to transmit their own information.In an ambient backscatter communication system,the wireless device can switch easily between energy-harvesting and signal-backscattering mode.The acquired energy is used to power device operations,such as circuit power consumption and signal sensing operations.This technology is very attractive,it can overcome the technical limitations of traditional active and passive tags in the Internet of Things to some extent,enabling lightweight computing communication devices to be embedded in everyday products at low cost,with a wide range in the Internet of Things era Application prospects.Generally,the radio frequency source and wireless channel have uncertainties.And the intensity of the carrier signal in the environment perceived by the backscatter device is uncontrollable and has a strong randomness,which also greatly affects the quality of ambient backscatter communication.In this paper,we build a model of ambient backscatter communication system.We model the mode selection problem of the tag as an infinite-horizon Markov Decision Process.And the bit error rate at the receiving end is deduced in detail.Given the channel distribution,this paper proposes an optimal mode strategy by value iterative algorithm,which achieves the communication rate performance as a theoretical limit.When the channel distribution cannot be accurately estimated,the Q learning method proposed in this paper can explore the sub-optimal strategy through repeated interaction with the environment.Finally,our simulation results show that the Q learning algorithm can continuously approximate the theoretical communication rate limit of the value iteration algorithm,and the performance of these two algorithms is significantly better than the other representative benchmark methods.