Optimized resource allocation for MIMO multi-carrier multi-user communication systems

In recent years, techniques of multi-user communication systems have been developed and significant performance improvements have been shown. For instance, in a Very-high-speed Digital Subscriber Line (VDSL) with Dynamic Spectrum Management (DSM) Level 3 capability, dramatic data rate increases and power savings are demonstrated. Performance enhancements are observed in multi-user wireless communication systems as well. The key to these improvements is the coordination among users, achieved by interference cancellation/avoidance together with proper resource allocation over the time, frequency, and/or spatial dimensions. However, current research works rarely consider the computational complexity, practical limitations, and the system operator's design objectives. Therefore, the gap between practical implementations and information-theoretical results has not been narrowed.; This thesis presents novel resource-allocation techniques for MIMO multi-carrier multiple-access channels (MAC) and broadcast channels (BC). To lower the computational complexity, zero-forcing generalized decision feedback equalizers (ZF-GDFE) are used at the receiver as the interference canceller. Starting from SIMO multi-carrier MAC, a weighted sum-rate maximization problem is formulated to characterize the achievable rate region. This problem can be solved in the dual domain efficiently by the Lagrange dual decomposition and a sub-gradient search. To evaluate the Lagrange dual function, independent tonal subproblems need to be solved. The solution to each of the tonal sub-problems requires finding the decoding order for the associated subchannel. Thus, an optimal tonal exhaustive search (TES) and two low-complexity suboptimal algorithms, successive ordering search (SOS) and modified greedy algorithm (MGA), are proposed to solve the decoding order.; Since rate maximization is not always the design objective of system operators, other optimization problems such as power control, proportional rate allocation, and margin maximization, are formulated. Interestingly, after the Lagrange dual decomposition, all the problems have the same tonal subproblem so that the techniques in weighted sum-rate maximization problems can be applied.; In MIMO multi-carrier systems, the precoding matrices of the MAC or the decoding matrices of the BC need to be considered as design parameters. To further generalize the algorithms to MIMO multi-carrier MACs and BCs, a singular value decomposition (SVD) and an SVD with projection (PSVD) algorithms for finding the precoding/decoding matrices are proposed. The resource-allocation problems for MIMO-MAC can then he solved by the proposed algorithms for SIMO-MAC with slight modifications. For MIMO-BC, the MAC-BC duality is applied so that the optimization problems can be solved by same techniques as are used for the MAC.; The performance of the proposed algorithms is examined by numerical simulations in VDSL and wireless communication systems. Simulation results show that all the optimization problems can be solved near-optimally in polynomial time. It is further demonstrated that the capacity region and the power region of Gaussian SIMO multi-carrier MACs with block fading can be closely approached via proposed efficient algorithms.