Massive Random Access Schemes Based On Reed-Muller Sequences

With the vigorous development of the Internet-of-Things(IoT),a wide range of advanced applications,such as smart city,smart traffic,and e-health,have emerged,leading to a surge in the number of IoT devices.Against this background,the third-generation partnership project(3GPP)has identified the use case of massive machine-type communications(m MTC),aiming to provide efficient and reliable connectivity for an enormous number of machine-type devices(MTDs).The random access(RA)procedure is the key to achieving this goal.However,the previous cellular systems are oriented to human-to-human communications and fail to meet the particular requirements of m MTC,especially massive connectivity,low signaling overhead,and low energy consumption.In this dissertation,we investigate the massive RA system design exploiting Reed-Muller(RM)sequences.The contents of this dissertation are summarized as follows.Firstly,we propose a joint active user detection and channel estimation algorithm based on RM sequences.Because of the vast sequence space and the computationally-efficient detection algorithms,RM sequences are well suited to serve as preambles in grant-free RA schemes.However,the existing detection algorithms exhibit poor robustness against the increasing number of active users.Hence,it is necessary to study how to further improve the capability of detecting RM sequences.We find that there is an elegant nested structure between the subsequences of RM sequences.On this basis,we design a layer-by-layer RM detection algorithm in the single-sequence scenario,which exhibits superior detection performance and channel estimation accuracy.For mitigating the potential error propagation,we further conceive an enhanced algorithm with moderate complexity.Besides,an iterative RM detection and channel estimation algorithm is proposed for the multi-sequence scenario.On the one hand,the impact of wrongly detected sequences is reduced with iterations.On the other hand,the detection capability and channel estimation accuracy can reinforce each other owing to the bi-directional information exchange.The successful detection probabilities of the above algorithms are analyzed theoretically.Our simulation results verify that the proposed algorithms are superior in terms of detection capability,channel estimation accuracy,and computational complexity.Secondly,we design an incremental RA scheme exploiting the nested structure of RM sequences.Since the traffic pattern in m MTC is typically short and burst,the network state is highly dynamic.To provide reliable and instant access in this case,RA schemes are required to handle potential access failures effectively.The commonly used retransmission scheme is liable to cause system congestion and increase access latency,and hence we conceive a more efficient solution in this dissertation.Specifically,we design an expansion rule according to the nested structure of RM sequences.The active users who suffer from access failures have to expand their RM sequences following this rule before sending them to the access point(AP).The AP executes the proposed recursive detection algorithm,where the expanded RM sequences are combined with the previously transmitted ones for attaining potentially better detection capability.Besides,by utilizing the discrepancy in the sequence length,the recursive detection algorithm can separate different batches of active users and detect them progressively,thus reducing multi-user interference.According to our simulation results,the proposed incremental RA scheme can improve the successful detection probability with low access latency.Thirdly,an unsourced massive RA scheme is proposed based on RM sequences.In unsourced random access(URA),the AP is only interested in the transmitted messages,regardless of their sources.Under the packetized and slotted transmission framework,the URA procedure can be partitioned into the slot-based information transmission and the message stitching among different slots.Tree coding is widely used for realizing message stitching,which,however,has the disadvantage of low spectral efficiency and high decoding complexity.In this dissertation,we propose a novel URA scheme exploiting the structural properties of RM sequences.Considering that the vast RM sequence space can improve spectral efficiency and the nested structure facilitates efficient information recovery,we utilize RM sequences to carry information chunks in slots.Besides,we explore the shift property of RM sequences and exploit it to design sparse shift patterns(SSPs).In addition to carrying partial information bits,SSPs also serve as the hints of message stitching.The influence factors concerning the slot-based RM detection are characterized,and the complexity of the SSP-based message stitching is analyzed.Simulation results validate the distinct advantage of our RM-based URA scheme in error probability and computational efficiency.