Pilot Contaimination Suppression And Spaitial Modulation In Massive MIMO Systems

The demand of social and economic development and technological advances are forcing wireless communications to undergo a transformation every decade,i.e.,creating a new standard.We are now in the research phase of rolling out the fifth generation of mobile communication technology(5G).The commercial deployment of 5G has also been accelerated around the world,with industry and academia looking beyond 5G and conceptualizing 6G.The work of this thesis was started when 5G was in full swing,hoping to make meager contribution to the current and the next generation of wireless communications.No matter what stage wireless communication is in,one of the main purpose is to increase system capacity close to shannon's limits.To achieve this goal,there are usually three ways:the first is to use ultra-dense network architecture,such as small-cell technology;The second is to move to higher frequencies and expand bandwidth,such as millimeter wave technology(mmWave);The third is to use technologies that have higher spectral efficiency.The multiple-input multiple-output(MIMO)technology studied in this thesis is one of the technologies to improve spectral efficiency.Massive MIMO technology refers to equipping hundreds of antennas at the terminal side,greatly improves the spectral efficiency and energy efficiency with in-terference reduction through spatial multiplexing and array gain.Moreover,along with the increasing of antennas,the random channel vectors between the users and base station become nearly orthogonal,which makes that noise and interference in the use of simple linear processing conditions can be eliminated.In addition,Mas-sive MIMO system has many other excellent features,such as greatly enhancing the robustness of the system,and significantly reducing the time delay in the commu-nication system.In order to realize the benefits of Massive MIMO,the base station needs to know fairly accurate uplink and downlink channel state information(CSI),so that effective precoding and signal detection can be carried out.In massive MI-MO system,uplink channel estimation is done by users sending pilot sequences to the base station.Then,the base station use the pilots for the estimation of uplink channel.But for the downlink channel estimation.typical method will not work.More precisely,cach base station transmits pilots to users side and the users esti-mate the downlink channel and then feedback to the base station side.Because the channel remains constant only within a certain period of limited time,i.e.,the chan-nel coherence time,and there are a large number of antennas at the base side,pilots will take up excessive time resources which results in inefficient data transmission Therefore.TDD duplex mode is usually adopted in massive MIMO system,so that the reciprocity of channels can be used to estimate downlink channels.However there is a serious problem in the channel estimation process,called pilot contami-nation.Pilot contamination is caused by using the same set of pilot sequences or related pilot sequence in the neighboring cells,thus the estimated channel of the base station to its own user inevitably contains the channels to users with the same pilot or related pilot sequence in other cells,and so also can send or receive interfer-ence signals from other cells.Furthermore,even when the antenna number tends to infinity,the interference caused by pilot contamation has no tendency to decrease which has been a bottleneck problem in massive MIMO system.Therefore,how to solve the pilot contamination problem and how to study the influence of pilot con-tamination on system performance are important for massive MIMO.In addition spatial modulation(SM)is a special MIMO technique.It requires only one radio frequency(RF)link to active one antenna among multiple antennas to send infor-mation,which can reduce hardware loss and avoid inter-antenna interference(IAI)and synchronization problems.The working principle of SM is that one part of the information source is used to select the antenna,and the other part is used for tra-ditional amplitude/phase modulation(APM),which can achieve a certain spectral efficiency gain.Hence,SM technology can be applied to the user side of massive MIMO systems to further increase the spectral efficiency of the system.General-ized Spatial Modulation(GSM)can stimulate multiple antennas simultaneously to send different data strems.Compared with SM,GSM can obtain a higher spectral efficiency gain at the cost of increased RF link and detection complexity,which can make a better trade-off in terms of spectral efficiency and system overheadThis thesis mainly focuses on how to eliminate pilot contamination and the performance under pilot contamination in massive MIMO system.Firstly,a power control method for reducing pilot contamination in massive MIMO svstem is pro-posed,and then a multi-cell minimum mean-square error(M-MMSE)precoder is deduced,which can eliminate the pilot contamination for downlink transmission under the condition that the channel covariance matrix of users using the same pi-lot is linearly independent from each other.In order to be more practical,a new channel covariance matrix estimation method is proposed,and the influence of pi-lot contamination on massive MIMO system is analyzed under imperfect channel covariance matrix information.Then the covariance estimation method is extended to analyze the spectral efficiency of the downlink in massive MIMO system with the massive uniform planar array(UPA)at the base station side under the same chan-nel condition.Finally,the uplink spectral efficiency of multi-cell multi-user massive GSM-MIMO system is analysed.The main research results of this thesis are as follows1.Two uplink pilot power control schemes in multi-cell massive MIMO sy system were proposed.The process of identifying the optimal pilot power of each user in-volved a combination optimization problem which was settled by genetic algorithm through mapping the pilot power of each user as an optimization variable into the information bit for iterative optimization.These two schemes could maximize the minimum signal-to-interference-plus-noise ratio(SINR)of the worst user and the mean SINR of users in the whole network for the downlink data transmission sepa-rately.In general,a constant power control mechanism is used,that is,each user is allocated the same value of transmitted power.The simulation results showed that our proposed maximized minimum SINR and maximized mean SINR power control schemes could significantly outperform the constant power scheme in terms of SINR and power consumption2.A novel channel covariance matrix estimation method was proposed and the spectral efficiency of the uplink and downlink of massive MIMO systems was ana-lyzed under imperfect CSI and pilot contaminition.Given the known information that massive MIMO can achieve unbounded spectral efficiency and climinate pilot contamination as long as the channel covariance matrices of users using the same pilot were linearly independent and the multi-cell minimum mean squared error(M-MMSE)receiver was adopted in the uplink.A M-MMSE downlink precoder under the same condition was derived and it was found that the spectral efficiency for the downlink also went unbounded without the effect of pilot contamination as the number of base stations antennas tended to infinity.However,the assumption of pre-known chanmel second-order information was not practical.Based on this in-vestigation.a novel method for channel covariance matrix estimation was proposed,which can achieve better performance compared to the existing second-order statis-tics estimation method.Furthermore.based on the imperfect covariance matrix obtained through the above methods,the spectral eficiency with different precoder-s or detectors was analysed and compared to the spectral efficiency with perfect covariance matrix both for uplink data transmission and downlink data transmis-sion3.A channel covariance estimation method which can be applied to two di-mensional antenna array was proposed and the spectral efficiency of downlink was analysed when massive uniform plane array(UPA)antenna array was adopted at the base station side under the condition of linear independence between covariance matrices of the co-channel users,which was beneficial to suppress pilot contamina-tion.Since the channel covariance method mentioned above can only be applied to one-dimensional antenna structure,namely uniform linear array(ULA),in order to remedy this defect,this method was extended to more general antenna layout and the UPA was analysed emphatically.As long as the antenna layout information is known at the base station side,the covariance value between antenna pairs can be estimated by using the characteristic that the covariance matrix does not have full degrees of freedom.After obtaining the estimation of the covariance matrix for UPA,it was applied to the multi-cell massive MIMO system and the downlink spectral efficiency was analysed with comparison with the existing method and the perfect covariance scenario under pilot contamination.Pilot contamination can be eliminated when the number of pilot sample is large enough.4.Generalized spatial modulation(GSM)was applied to the user side and combined with massive MIMO to form a massive GSM-MIMO system.Then the spectral efficiency of uplink in multi-cell multi-user massive GSM-MIMO system with pilot contamination was analyzed.So far,there is a paucity of literature on massive GSM-MIMO in a multi-cell multi-user system and some defects were found,a practical channel model having an arbitrary transmit correlation matrix and receive correlation matrix for each user and BS was proposed to analyse the system capacity.Pilot contamination was also considered.Single-cell MMSE(S-MMSE)was firstly used for detecting the classic APM signals as well as the index of transmit antennas.An algorithm was also developed for calculating the transmit antenna detection error probability using the order statistics of independent but non-identically distributed random variables.At the same time,the approximate sum-rate expressions of S-MMSE,zero-forcing(ZF)and match filter(MF)were derived.