Research On Active Device Detection And Channel Estimation Of Ambient Backscatter Systems

Targeting at the growing market of IoT,5G mobile technology takes massive machine-type communications(mMTC)as one of the three main scenarios.This kind of communication has the characteristics of large scale,low activity and low power consumption.The key challenge that mMTC faces is to detect active devices and to decode the data sent by the device in an effective and timely way.This thesis studies how to detect the activity of a large number of devices and to estimate the effective channel in an ambient backscatter communication(AmBC)system with a low latency.This thesis studies the problem of detecting the activity of massive number of high rate backscatter devices in an AmBC system and estimating the channel simultaneously.This thesis uses the grant-free access scheme to solve this problem.Compared with the traditional grant-based random access scheme,the grant-free access scheme we used here reduces transmission delay and signaling overhead,besides,it increases the number of devices that can be accommodated in the network.Further more,utilizing the sparsity of the effective channel,this thesis uses approximate message passing(AMP)algorithm to solve this problem.This algorithm is an efficient iterative threshold algorithm with a low computational complexity for compressive sensing.The performance of channel estimation is close to that of traditional minimum mean square error(MMSE)detection.Simulation results prove that this thesis can accurately detect the activity of a massive number of devices in AmBC systems and estimate their channels with a low latency.This thesis also studies the case where the data rate of backscatter devices in AmBC system is low,which detects the activity of multiple backscatter devices and estimates the corresponding channels.The elements of the perception matrix in the established model are not independent distributed,thus,the problem cannot be solved effectively by the AMP algorithm.To overcome this challenge,this thesis utilizes the characteristics of the radio frequency source symbol to modify the AMP algorithm,which eliminates the correlation between the elements of the perception matrix and introduces the spreading gain to improve the detection signal-to-noise ratio.It can be known from the simulation results that the modified AMP algorithm used in this thesis can detect the activity of multiple low rate devices and estimate the effective channel in the AmBC system.In particular,when the rate of the radio frequency source is constant,the faster rate of backscatter device,the better performance of detection and estimation.In this thesis,the AmBC system with a low cost and a low energy consumption is successfully incorporated into mMTC.In the scenario where multiple devices in the AmBC system communicate with each other at the same time,detailed architecture analysis and algorithm solutions are performed for different situations where the data rates of the backscatter devices are different.From the theoretical analysis and numerical results,it can be proved that this thesis has realized the activity detection and channel estimation for a large number of devices in the AmBC system and achieved a good performance.