| Machine-to-machine(M2M)communication,which aims at sensing and data collec-tion,is growing rapidly in scenarios of smart city,environment monitoring,smart agricul-ture,forest fire prevention,etc.,which gives rise to the requirements to support massive connections.Specifically,in the massive machine-type communication scenario,the fu-ture fifth generation(5G)network is required to support over 100 billion connections and10~6/km~2connection density.In this scenario,a vast number of users would attempt to access the network simultaneously.Frequent transmission collisions would lead to net-work congestion,unexpected delay,and radio resource wastage.The rational utilization of limited resources to support reliable access and efficient data transmission proves to be a huge challenge.This thesis studies the non-orthogonal multiple access protocols to solve the problem of access and data transmission of massive users.The main work and contributions are summarized as follows:(1)A novel non-orthogonal random access(NORA)mechanism,which combines non-orthogonal multiple access(NOMA)and long-term evolution(LTE)random access process,is proposed to alleviate collision and reduce the delay of random access.NORA utilizes the spatial distribution characteristics of users to design the arrival time-based multi-preamble detection and distance-based random access response(RAR)reception schemes.The RAR message format is tailored for practical realizations.As a result,NORA facilitates the simultaneous transmission of access request messages of multiple users on the same resource block using power-domain multiplexing instead of conduct-ing preamble retransmissions.Then the base station(BS)separate users'access request messages using successive interference cancellation(SIC),which fulfils the simultaneous access of multiple users.An analytical model is derived to investigate the performance of NORA,which includes throughput of random access(maximum number of supported users),average access success probability,average access delay,etc.Simulation results show that the NORA scheme can increase the maximum number of supported users by more than 30%compared to LTE.The average access delay for the successfully accessed users are reduced up to 50%.(2)The performance analysis and design of NORA are carried out in the large-scale network scenarios,which is motivated by the dense deployment of BSs with the dramatic increase of M2M users.Firstly,taking into account the constraints on the distances be-tween NORA users and their serving BS,the spatial distribution model of NORA users is obtained using stochastic geometry theory,based on which the inter-cell interference model is derived.Secondly,considering the characteristics of non-orthogonal multiple ac-cess,two power control schemes are proposed,namely partial compensation for the path loss with power back-off and overcompensation with power lifting.The analytical model of NORA in the large-scale network scenarios is then derived,which includes through-put of random access,average access success probability,NORA gain,etc.Simulation results show that partial compensation for the path loss improves fairness of NORA users as well as the NORA throughput.In addition,for a given multipath channel transmission characteristic,the NORA gain can be maximized by optimizing the BS density.(3)The coded random access(CRA)mechanism is studied and designed aiming at the requirements of the random access process of massive users.Users randomly select multiple slots to transmit access request messages repeatedly based on a predefined access strategy,forming the access pattern.The multi-user joint detection scheme is proposed based on SIC across slots and intra-slot SIC,which enables the reception of users'access request messages at the BS.Based on the AND-OR analysis method,an iterative model of the access request message detection probability is established taking into account the capture effect of intra-slot SIC.An analytical model is then derived to investigate the performance of CRA,which includes throughput of random access,average access failure probability,average access delay,etc.Simulation results show that CRA can double the maximum number of supported users compared with NORA.Furthermore,an analytical model of CRA is established for the large-scale network scenarios.Simulation results show that the maximum number of supported users of CRA in the large-scale network scenarios is reduced by about 50%compared with that in single-cell scenarios.Moreover,under the same system load,the user access strategy that achieves the lowest access failure probability requires lower average repetition rate in the large-scale network scenarios than in the single-cell scenarios.(4)Aiming at the user asynchronous scenarios,the multi-domain non-orthogonal multiple access schemes,which jointly considers power-domain,code-domain and time-domain,are designed.On the one hand,for the symbol asynchronous scenarios of NOMA users,a time-domain and power-domain joint non-orthogonal scheme,namely asynchronous NOMA(ANOMA),is designed in combination with the over-sampling and belief prop-agation techniques.The performance of the ANOMA system is analyzed in terms of sum data rate and bit error probability.Simulation results show that the sum data rate of ANOMA exhibits noticeable gain over synchronous NOMA.In addition,ANOMA using belief propagation detection shows similar performance to that of single-user transmission in terms of bit error probability.On the other hand,a time-domain and code-domain joint non-orthogonal scheme,namely asynchronous CRA(ACRA),is proposed using time di-versity and combining techniques for the slot asynchronous scenarios of CRA users.The performance of the ACRA system is analyzed in terms of throughput and average access failure probability.Simulation results show that CRA with maximum ratio combining(MRC)can double the maximum number of supported users compared with slot syn-chronous CRA while reducing the average access failure probability by at least an order of magnitude. |
PDF Full Text Request