Research On Random Access Schemes For Massive Machine-Type Communication Systems

With the technological advances in wireless communications,the services provided by mobile communication systems have evolved from simple voice calls in 1G to diversified data services in 5G.Furthermore,driven by the social demands,the Internet of Things(Io T),which enables information exchange among physical objects,exhibits great application potentials and positive market prospects.The massive Machine-Type Communications(mMTC)is a key application scenario in 5G standard to support the Machine-to-Machine(M2M)communications for Io T.Furthermore,the devices in mMTC systems exhibit the following remarkable features,i.e.(1)massiveness and ultra-high deployment density,(2)the demand for low cost and low power consumption,and(3)low activation probability with small data packets.These features have brought new challenges to existing random access(RA)mechanisms in mobile communication systems.Specifically,the massiveness of devices causes the shortage of RA resources,which results in RA collisions and wastage of data-transmission resources.Moreover,a handshaking procedure is required between activated devices and the base station(BS)to exchange the control signals.However,this handshaking procedure undermines the transmission efficiency for small data packets,and increases the power consumption for low-cost devices.In order to address the challenges encountered in mMTC systems,this paper considers different mMTC scenarios,and studies the design of grant-free RA schemes,as well as the detection algorithms at the receiver.Specific contents of this paper are described as follows.Firstly,for the mMTC systems in static scenarios,a fixed-symbol aided RA scheme is proposed to address the user-activity detection(UAD)problem and the RA collision problem.In the proposed RA scheme,each activated device simply inserts a fixed symbol into its data packet before transmission.According to the uplink channel state information(which is assumed known to the BS in static scenarios)and the received signals of the fixed symbols,a Message-Passing Based Activity Detection(MP-AD)algorithm is proposed.A factor graph is established for the MP-AD algorithm,and the message updating rules are derived for different nodes on the factor graph.According to the UAD result from the MP-AD algorithm,the multi-antenna BS can perform parallel multi-user detection to recover the data from collided devices.To alleviate the correlation problem of the MP-AD algorithm,a Deep Neural Network Aided MP-AD(DNN-MP-AD)algorithm is proposed,where the iterative message passing process on the factor graph is transformed into a forward-propagation message passing process in a DNN.The simulation results are presented to show the throughput performance of the proposed fixed-symbol aided RA scheme,as well as the detection accuracy of the MP-AD algorithm and the DNN-MP-AD algortithm.Secondly,for the crowded mMTC systems in non-static scenarios,a DNN-Aided Message Passing-Based Block Sparse Bayesian Learning(DNN-MP-BSBL)algortihm is proposed to alleviate the convergence problem of existing message passing-based detection algorithms.In this algorithm,the edge-type message passing process on a factor graph is transformed into a node-type one in a DNN.Weighting parameters are imposed on the messages passed in the DNN,and these parameters are further trained to mitigate the correlation problem of the messages.The simulation results show that the DNN-MP-BSBL algorithm could effectively alleviate the convergence problem,thus improves the UAD and channel estimation(CE)accuracy for crowded mMTC systems.Thirdly,for mMTC systems based on serial multi-user detection,a random Non-Orthogonal Multiple Access(NOMA)transmission scheme is proposed,where uplink power allocation is avoided to meet the requirements of the low-cost and low energy consumption devices.To improve the packet-recovery probability,a cross-slot successive interference cancellation(SIC)packet recovery scheme is proposed,where the intra-slot SIC decoding and cross-slot interference cancellation are performed iteratively to recover data packets.The cross-slot SIC packet-recovery procedure is modeled as a Markov process,based on which a throughout analysis is derived.Finally,the sum rate of the proposed random NOMA transmission scheme is maximized by jointly optimizing the transmission probability and encoding rate of the devices.Finally,for mMTC systems in low-earth orbit(LEO)satellite-enabled Io T,a terrestrialsatellite grant-free RA scheme is proposed,which is designed for the high mobility of LEO satellites and the land-mobile satellite channel.In this terrestrial-satellite grant-free RA scheme,a Bernoulli-Rician Message Passing with Expectation Maximization(BR-MP-EM)algorithm is proposed for the joint UAD and CE problem.The BR-MP-EM algorithm is divided into two stages.In the inner iterations,the Bernoulli messages and Rician messages are updated for the joint UAD and CE problem.Based on the output of the inner iterations,the expectation maximization(EM)method is employed in the outer iterations to update the hyper-parameters related to the satellite channel.Finally,simulation results show the UAD and CE accuracy,as well as the robustness of the proposed BR-MP-EM algorithm against the satellite channel hyper-parameters.