| Mobile communication systems have evoled from conventional low-capacity analog mode to digital mode nowadays. As the number of users grows rapidly and new-type wireless businesses emerge, the transmission rate and spectral efficiency are required to grow accordingly. For the whole communication networks, conventional processing method in a single cell is doomed to be limited by inter-cell interference and can not enhance the performance of the entire networks. In recent years, multi-cell cooperative processing is proposed as a concept to boost the spectral efficiency of communication networks.Multi-cell cooperative processing refers to a class of technologies. Using cooperative transmission or detection through multiple access points, multi-cell cooperative processing technologies can effectively exploit or mitigate inter-cell interference. Specifically, cooperating BSs share CSI or even user data to eliminate or exploit interference. Similar to multi-cell cooperative processing, the distributed architectures in LTE-A have been proposed and accepted by the 3GPP organization. However, the capability of the interface between BSs is relatively low, and cooperation is confined among limited BSs. In other words, multi-cell cooperative processing is potential to be further exploited to increase the performance of the whole communication networks.So far, the research on multi-cell cooperative processing has accomplished some achievements world-widely. On uplink, distributed detection algorithms are focused on. To lower the complexity of detection algorithms, this class of algorithms models a whole network into a factor graph, and uses message passing algorithms to detect user symbols on the factor graph. Those distributed algorithms focus on complexity, cost on message passing and quantization of passing messages. To trade off complexity and performance, networks can be partitioned into clusters within which distributed detection is performed.On downlink, the research is concentrated on multi-cell beamformer design. Multi-cell beamformer design aiming to maximize the Shannon capacity of a multi-cell network is proved to be an NP-hard problem. Various suboptimal algorithms are proposed to achieve high quality solutions. Distributed implementation for large scale networks is also an important research direction.This dissertation studies serveral problems of iterative interference cancellation on uplink and multi-cell beamformer design on downlink. Three research points are included as follows.1. This dissertation studies the achievable rate region of iterative interference cancellation in multi-cell networks on uplink. Methodologically, nested lattice code is for the first time used as a theoretical tool to study the achievable rate region of iterative interference cancellation. Nested lattice code has structures and achieves the Shannon capacity of AWGN channel, so it is theoretically advanced and potential for practical realization. With the excellent theoretical properties of lattice, the lower and upper bound of rate region on iterative interference cancellation for both two-cell network and linear Wyner-modeled network. In addition, this dissertation extends to general multi-cell networks, and obtains the convergence condition of iterative interference cancellation and minimum achievable rate of a single user. An important conclusion is drawn that the asymptotic achievable rate can approach the single-cell interference-free scenario provided the convergence condition is satisfied.2. From the practical point of view, this dissertation proposes a low-complexity and MMSE-based iterative interference cancellation scheme for massive MIMO multi-cell networks. The key idea is that inter-cell interference is directly estimated using the MMSE estimation method, and then exchanged for interference cancellation to increase the estimation accuracy of interference for the next iteration. Following this, the dissertation obtains the general expression of the interference estimates and covariance matrices of estimation errors of the proposed iterative interference cancellation scheme. The general expression of covariance matrices of estimation errors can be used to analyze the convergence of iterative interference cancellation in any networks. The simulation shows that the theoretical results fit well with the experimental data on the convergence behavior. Considered that some symbol estimates get high confidence level in the iterative process, this dissertation proposes to make hard decisions on these symbols before interference cancellation. It is shown in the simulation that the detection performance can be boosted gradually by several iterations, and hard decisions on symbol estimates with high confidence levels can further shorten the performance gap to the interference-free condition.3. This dissertation proposes a unified framework of parallel and distributed optimization algorithm for partial cooperation in both HetNet and massive MIMO multi-cell networks. A user-centric partial cooperation strategy is proposed to strike a better compromise between high spectral efficiency and cooperation costs. To increase the real system performance, sum rate is targeted to model the optimization problem. The dissertation obtains an equivalent problem, which is easier to solve than the former one. The overcome the coupling problem brought by the user-centric partial cooperation, the framework decouples the equivalent optimization problem using layered decomposition technology. The problem on the upper layer has a distributive form, and the subproblems on the lower layer are independent. In addition, the dissertation takes advantage of partial separability of objective functions in subproblems and proposes a parallel algorithm for subproblems. The simulation shows that the framework of optimization algorithm can obtain high-quality beamformers and the parallel optimization of subproblems can greatly decrease the number of iterations. For massive MIMO multi-cell networks, the framework is significantly advanced in power consumption.
|
PDF Full Text Request