On The Design Of Low-complexity Massive Access Schemes

Internet-of-Things(IoT)is one of the key applications in 5G.In the IoT scenarios,numerous always-on-line devices require the future wireless communication system to support massive connections with guaranteed data rates.Furthermore,with the rapid development of wireless networks,the network state,the traffic state and the channel state,become more dynamic,which requires the system to be highly adaptive.All these impose huge impacts on the access protocol design.In this dissertation,resorting to information theory and coding theory,we investigate the access protocol design in the massive access scenarios.The contents of this dissertation are summarized as follows.Firstly,a novel random non-orthogonal multiple access framework called Rateless Multiple Access(RMA)is established.In the framework,instead of granting active users with specific resources in a fixed and centralized manner as before,the access point simply assigns a random access control function,with which they share a resource block in a pseudo-random manner.The transmission process of each user,as well as those of all the users as a whole,resembles a special form of linear superposition rateless encoder and the low-complexity belief propaga-tion algorithm is used to retrieve the information at the access point.For further enhancing the performance,we apply the idea of spatial coupling to RMA and propose the enhanced RMA in which all the user codewords are properly spatially-coupled among different subsets of re-source elements.Based on the extrinsic information transfer theory,we theoretically analyze the relationship between the achievable throughput and the random access control function,spatial coupling parameters and precoders,based on which,we propose a method to jointly optimize all the key system parameters.According to the analytical and simulation results,RMA asymptot-ically approaches the channel sum capacity with high adaptability and low signalling overhead,which makes it a viable candidate for future massive access.Secondly,by exploiting the intrinsic access pattern of each user in the RMA framework,we propose a grant-free RMA scheme,which no longer needs the registration procedure as in the original RMA,thus further reduces the signalling overhead and system latency.Furthermore,a low-complexity joint iterative detection and decoding algorithm is proposed,with which,the channel estimation,active user detection,and information decoding may be done simultaneously.Then,a method based on density evolution is proposed to evaluate the system performance.Analytical and simulation results are consistent in demonstrating that the grant-free RMA only has a very little sacrifice in the block error rate performance while greatly reducing the signalling overhead and system latency.At last,the idea is extended to the grant-free sparse code multiple access(SCMA),based which,a scheme of SCMA based on signature hopping is proposed.Thirdly,to simplify user paring in uplink non-orthogonal multiple access(NOMA),a novel rate-splitting NOMA scheme is proposed.In the scheme,by splitting its information into mul-tiple flows,each user may create multiple virtual users that can be then successively decoded even with the same arriving power,which avoids the need of user paring and thus reduces the scheduling complexity.Furthermore,we analyze the performance of the scheme,and derive exact closed form expressions for outage performance and achievable sum rate.Moreover,a method to design key parameters,i.e.,rate splitting pattern and power splitting pattern,to op-timize the performance is proposed.Finally,analytical and simulation results are consistent in demonstrating that the proposed rate-splitting NOMA achieves distinctively higher sum rates and lower outage probabilities than previous schemes.