The Study Of Signal Detection Method For Ambient Backscatter Communication System

Ambient Backscatter Communication(Am BC)is an important wireless communication method with high spectral efficiency and low energy consumption.To avoid the occupation of new spectrum resources and reduce the transmitting energy by passive transmission,Am BC uses the existing signal in the surrounding environment as the carrier and loads information through the secondary modulation of the environmental electromagnetic signal in the radio frequency domain.Am BC is a key technology to achieve wide-area and low-energy wireless network coverage.However,the detection of Am BC signals under complex non-cooperative reception conditions is very challenging due to the dynamic fluctuation characteristics of ambient electromagnetic signals and interference to receivers.Therefore,it is a key technology to achieve high-performance Am BC.At the same time,the large deployments of Am BC system will give full play to the spectral efficiency advantages,but will also bring significant self-interference.Its achievable coverage performance and spatial spectral efficiency performance need to be further studied.Based on the theory of statistical Bayes,deep reinforcement learning and stochastic geometric wireless network modeling theory,this paper studies Am BC signal detection and the coverage of large-scale networks deeply from various aspects such as multiple antenna readers,link anti-interference,feature extraction and learning,signal detection prototype system construction and the coverage of the large-scale network.The main innovative works and achievements are summarized as follows:1.To overcome the high bit error rate of detection and low transmission rate in the traditional Am BC system,a bayesian-maximum likelihood estimation signal detection method with multi-antenna is proposed.Firstly,a multi-antenna Am BC system model is designed,which offers transmit-receive diversity.Secondly,an efficient Am BC signal detection method in a multi-antenna environment is developed based on Bayesian optimization and maximum likelihood estimation theory.Finally,an optimal detection threshold selection scheme is designed based on the non-central chi-square distribution to improve the detection performance.The simulation results show that the method has good detection performance,which owns a higher transmission rate,better detection accuracy and lower energy consumption.2.To solve the problem of incomplete parameter acquisition and serious direct link in-terference in traditional Am BC signal detection methods,a multi-antenna signal detection method based on a double-delay depth enhancement strategy is proposed,which realizes an efficient and accurate signal detection performance.Firstly,based on the programmable reconfigurable intelligent surface(RIS)and the multi-antenna Am BC architecture,a RIS assisted multi antenna Am BC system model is designed.The detection probability is improved by optimizing the RIS phase shift,and the detection problem is transformed into an optimization problem.Secondly,to solve the nonconvex optimization problem,the Twin Delayed Deep Deterministic Policy Gradient is adopted to solve the over estimation of Q value in the traditional deep reinforcement learning algorithm by taking the minimum value of Q value estimated by two groups of neural networks.The stability of the algorithm is improved by delay updating and smooth regularization of the target policy network.Thus,the optimal solutions of the nonconvex optimization problem and the lower bit error rate are obtained.Finally,the effect of a hyperparameter of the algorithm is explored and the hyperparameter is chosen appropriately to further improve detection performance.A large number of numerical simulation results show that the proposed intelligent detection method of backscattered signal in multi antenna environment obtains a lower bit error rate and more stable convergence performance than the reference method under the condition of lack of channel state information.3.Aiming at the coverage performance of the backscatter device-to-device(D2D)network in the large-scale environment,a two-dimensional large-scale environmental backscattering D2 D network model is constructed by using the random geometry method first.Then,the D2 D link transmission protocol is designed and the interference between D2 D devices is modeled and analyzed.Moreover,the network coverage probability and spatial spectrum efficiency are deduced theoretically.On this basis,Monte Carlo simulation is used to verify the theoretical analysis.The analysis and simulation results show that the network coverage increases significantly with the decrease of the reflected main lobe beam width.In addition,with the increase in node density,the regional spectral efficiency decreases gradually.Moreover,due to the small co-channel interference from the surrounding backscattering equipment,the narrower reflected beam width leads to better regional spectral efficiency.The transmission reliability of the Am BC system can be improved significantly by the above research results,while the reachable coverage and spectral efficiency performance of the large-scale Am BC network can also be fully evaluated.It owns important theoretical significance for improving the signal detection performance of the environmental backscatter link and realizing the wide area coverage of the low-power passive backscatter Internet of Things.