Non-Orthgonal Multiple Access Transmission Over Millimeter Wave Massive MIMO Multi-Cell Networks

The explosive growth of mobile traffic and mobile devices worldwide has made current wireless networks face greater challenges in terms of transmission rate,transmission delay,and network connection capability.Therefore,it is extremely urgent to seek a breakthrough in the physical layer to support more user access and provide higher transmission rates.In this context,millimeter wave(mm Wave)massive multiple-input-multiple-output(MIMO)technology emerges.With its abundant frequency bands,mm Wave is capable of providing network connection for huge devices as well as higher transmission rates.Moreover,combining with massive MIMO technology,the severe pass loss of mm Wave can be combated effecitively and great spatial divisity and multiplexing gains can be further achieved,resutling improved system spectral efficiency.In the mm Wave massive MIMO system,users can be coveraged by spatial beams through employing effective beamforming techiniques at base station(BS)side equipped with a large number of antennas,thereby realizing spatial resolution by using different beams to cover different users spatially.On this basis,different users can be served by differenct spatial beams without mutual interference over space.Nevertheless,a specific beam will cover more than one user under ultra dense network scenarios with massive users distributed,which leads to inter-user interference within each beam.Hence,researchers propose power-domain non-orthogoanl multiple access(NOMA)to deal with aforementioned problem.With NOMA,system spectral efficiency can be further improved due to beam resources sharing by multiple users.Currently,various schemes with respect to mm Wave massive MIMO-NOMA have been proposed though,there are still some challenges in terms of channel estimation,precoding/decoding design and transmission design over multi-cell scenarios.To overcome those challenges,in this dissertation,we propose both cooperative and distributed multi-cell mm Wave massive MIMO-NOMA tranmission schemes.The proposed schemes are further optimized through maximizing system sum rate.Meanwhile,a more effective channel estimation method is proposed based on sparse Bayesian learning framework,whose design is a strong support for the correspoding NOMA transmission schemes.The specific contents of this dissertation are listed as follows:1.In view of the sampling mismatching for users' angle of arrivals of the channel estimation schemes with discret Fourier transformation basis,we propose an effective channel estimation approach for mm Wave massive MIMO systems based on off-grid model,where the corrsponding angle sampling mismatching is modeled as a mismatch error vector.As a result,the better estimation performance with less pilot training sequences has been achieved through emoloying effective solving methods.To be specific,we first formulate a parameterized off-grid mm Wave massive MIMO channel estimation model involving a mismatch error vector to describe the sampling mismatching for angles.Then,the whole channel estiamtion can be divided into two parts: learning of channel statistics and simultaneous virtual channel estiamtion.We propose expectation maximization based sparse Bayesian learning approach to learn channel statistics and further group users into difference clusters according to the learned parameters to reduce pilot training sequences.In addiction,we resort to the linear minimum mean square error method to estimate the instantaneous virtual channel with less pilot overhead.Finally,we corroborate the validity of the proposed method through numerical simulations,showing better minimum mean square error performance in constrast to the current schemes.2.In view of the bottleneck of currrent mm Wave massive MIMO-NOMA strategies to address inter-cell interference as well as the lack of optimal tranceivers design,we propose a cooperative multi-cell mm Wave massive MIMO angle-domain NOMA transmission scheme.This scheme is optimized by maximizing system sum rate to pursue optimal precoding/decoding and user scheduling designs,with help of convex optimization methods as well as interference alignment technology.The main idea of angle-domain NOMA is to schedule one cell-center user and one cell-edge user along with the same spatial beam to conduct downlink NOMA transmission.We seek optimal precoding/decoding and user scheudling to maximize system sum rate,where precoding design is decomposed into outer and inner ones.We then design outer one with the help of users' beam signatures and propose two strategies for inner precoders and decoders design,i.e.,joint optimization of precoders/decoders(JOPD)and cooperative NOMA(C-NOMA).Specifically,in JOPD,optimal precoders/decoders are achieved through solving a nonconvex problem cooperatively within multiple cells,which is subject to users' quality of service(Qo S)to maximize system sum rate.An alternate optimization algorithm based on the constrained convaceconvex procedure is proposed for the solutions of the nonconvex problem.In C-NOMA,we obtain simplified precoders/decoders by employing interference alignment technology to eliminate inter-cell interference.Then the origianl optimization problem is reduced to a optimization problem related to power allocation,which is handled by an iterative algorithm.Finally,user scheduling strategies for both JOPD and C-NOMA are separately designed to further improve system performance.Simulations results show the efficiency of proposed algorithms,in addition to the better system performance as well as Qo S satisfaction of the proposed schemes in constrast to the current ones.3.In view of the needs of information exchanging between multiple cells and cooperative optimization within networks,we further propose a distributed multi-cell mm Wave massive MIMO angle-domain NOMA scheme,where the optimal transmission can be obtained through the optimization within each cell distributedly,subject to the interference threshold of each cell to the users located in the other cells.Consequently,there is no need to exchange information and solve optimization problems cooperatively within multiple cells.Meanwhile,we involve hybrid precoding framework in this scheme considering less ratio frequency chains equipped at the BS side to reduce the cost of practical deployment and energy consuming.In detail,the main idea of distributed angle-domain NOMA is to schedule two users over the same spatial beam to conduct downlink NOMA.This scheme is optimized through solving a nonconvex optimization problem to maximize system sum rate,subject to the interference threshold of each cell to the users in the other cells,users' Qo S and modulus constraint of analog precoding matrix.To solve the nonconvex optimization problem,we propose a two-step solving method,i.e.,seeking optimal analog precoding design at the first step and jointly optimizing digital precoders/decoders at the second step.Specifically,at the first step,though fully considering the characteristics to design the analog precoding matrix,we reformulate the problem with respect to the analog precoding to maximize spatial power for desired NOMA user clusters while minimizing interference for the other clusters,whose problem is finally established as a multiple objective optimization one.Then a weighted sum method is proposed for its solution and an efficient analog precoding design strategy is obtained correspondingly.At the second step,we propose an alternate optimization based on quadratic transform algorithm to transform the orignal problem regarding to the digital precoders/decoders into a convex problem,achieving a suboptimal solution.Compared to the current transmission scheme,the proposed one can obtain better system performance in terms of sum rate through dealing with inter-cell interference effectively.