Adaptive Transmission For Wireless MIMO Communications

Wireless multiple-input multiple-output (MIMO) systems that employ mul-tiple transmit antennas and multiple receive antennas are considered to be a key enabler for exploiting spatial resources and improving the system throughput, spectrum efficiency, power efficiency, and transmission reliability. During the last nearly twenty years, considerable efforts have been putted into developing signal processing algorithms to realize the potential of MIMO in practical implementa-tions. MIMO technique is now an inherent part of several emerging communica-tion standards. The future mobile communication systems are required to provide higher system throughput, spectrum efficiency, power efficiency, and transmis-sion reliability. The large-scale antenna configuration is necessitated. Based on the backgrounds, we investigate the theoretical methods of adaptive transmission for wireless MIMO communications, including link adaptation method for MIMO transmission with turbo receivers, and adaptation method for pilot and power allocation in massive MIMO systems.Firstly, we investigate the performance modeling of turbo receivers for adap-tive MIMO transmission. The turbo receivers are incorporated with link adap-tation schemes to achieve promising performance gains for MIMO transmission, in parallel, adapt the link in the actual channel conditions with assist from per-formance modeling. We propose the performance modeling in terms of equiva-lent model, transitive parameter, performance abstraction and transitive relation. Based on the Gaussian distribution of equivalent model for log-likelihood ratio, the proposed methods are to derive the transfer chart with mutual information as transitive parameter, to derive semi-analytical transitive relation with bit er-ror rate or fidelity as transitive parameter and to derive semi-analytical transitive relation with mutual information as transitive parameter. In so doing, the per-formance prediction methods for soft-input soft-output (SISO) components of the turbo receiver are proposed. Simulations based on broadband wireless commu-nication system, are conducted to validate the analytical result and to show the advantage of the proposed method.Then, an efficient link adaptation scheme with turbo interference cancela-tion receivers is proposed for MIMO transmission. We derive the post-signal-to-interference plus noise ratio (P-SINR) and effective signal-to-noise ratio (E-SNR) for turbo interference cancelation receivers. We determine the range of rank indi-cation and channel quality indication with classical link adaptation scheme. We propose the efficient link adaptation scheme, which reduces calculated amount of the link adaptation scheme by predicting online throughput only in the shrunk range of rank indication and channel quality indication. Simulations reveal that the proposed scheme gets more throughputs compared to the classical link adap-tation scheme and obtains 3.5dB gain. Meanwhile, the scheme is satisfied with the QoS (Quality of Services).Next, we investigate the achievable rates with minimum mean-square error (MMSE) receiver for massive MIMO systems, under the angular domain chan-nel representation. We decompose P-SINR into independent random variables, of which the first two moments and probability density functions are derived. The normalized P-SINR is therefore proposed to be Gamma distributed, and the prob-ability distribution function of P-SINR can be well approximated by a Gamma distribution. In this light, a lower bound on the P-SINR and approximation of achievable rate are derived. Numerical results demonstrate that both the lower bound on the P-SINR and the approximated rate apply for a finite number of antennas and remain tight.Finally, we adapt the pilot-assisted multi-user massive MIMO systems in the actual channel conditions with pilot and power allocation. We utilize the angular domain channel representation, and adopt linear MMSE channel estimation and detection. For tractable analysis and low-complexity solution, a tight approxi-mation on the achievable sum-rate is derived. For a given coherence interval and total energy budget, we study the joint optimization of the training length and the training power to maximize the approximate achievable sum-rate. The training length optimization for fixed training power and the training power optimization for fixed training length, are both shown to be concave. An alternative optimiza-tion that solves the training length and power iteratively is proposed for the joint optimization. In addition, for the special case that the training and data trans-mission powers are equal, we derive the optimal training lengths for both high and low SNR regions. Numerical results show the tightness of the derived sum-rate approximation and also the significant performance advantage of the proposed joint optimization.