Research On Detection Techniques Of Grant-Free Random Access Systems

Grant-free random access is a critical technology for the fifth generation(5G)mobile communication,as it effectively reduces signaling overhead,transmission delay,and power consumption by eliminating cumbersome signaling interactions used in traditional grantbased access schemes.However,existing research on signal detection for grant-free random access suffers from limitations such as low accuracy,small numbers of active connections,and low transmission rates.These challenges hinder the deployment of grant-free random access for diverse communication services and massive connected terminals.To address these limitations,this thesis researches signal detection techniques for grant-free random access to support a large number of active connections under diverse propagation environments such as Rayleigh,generalized,and fast time-varying channels.The main contributions are summarized as follows:1.To improve the detection accuracy and number of active users supported under Rayleigh channels,a near-optimal signal detector for grant-free random access systems is designed.An approximated average bit error rate probability(ABEP)of grant-free random access system,which is able to represents the bit error rate(BER)of the optimal detector is derived in a closed-form expression.Two near optimal detectors,residual oriented multipath orthogonal matching pursuit(ROM-OMP)and subspace argument matching pursuit(SAMP),are proposed to improve detection performance for the burst-sparsity and framewise joint sparsity models,respectively.Simulation results verify the correctness of the ABEPs and demonstrate the BER performance of the proposed detectors is capable of outperforming conventional schemes by 3 d B.2.To improve the generalization capacity of detection algorithms under different transmission environments,near-optimal detectors for grant-free random access systems under generalized channels are proposed.A closed-form expression of ABEP for grant-free random access systems under generalized channels is derived.Two extended alphabet-based expectation propagation(EA-EP)and block expectation propagation algorithms are proposed to improve the accuracy of active user detection.Simulation results verify the correctness of the derived ABEPs and demonstrate the BER performance of the proposed detectors outperforming conventional schemes by 2 d B.3.To ensure both transmission rate and reliability under fast time-varying channels,non-coherent signal detection techniques are researched.A non-coherent grant-free transmission scheme based on dual-mode generalized index modulation is proposed to improve the transmission rate.A low-complexity coordinate descent algorithm and dynamic threshold scheme are proposed for blind detection of active users,and an index-constrained hierarchical decoding algorithm is designed for data decoding.Simulation results demonstrate the advantages of the proposed non-coherent transmission scheme in improving transmission rate,enhancing BER performance,and reducing computational complexity.In summary,this thesis provides a theoretical basis and algorithm support for the development and application of grant-free random access signal detection technology in different application scenarios of the Internet of Things(Io Ts).It addresses limitations in current research and proposes novel solutions that improve detection accuracy,number of active users supported,generalization capacity,transmission rate,and reliability under different propagation environments.