Resource allocation for pilot-assisted massive MIMO transmission

Abstract

This paper is on the resource allocation problem for pilot-assisted multi-user massive multiple-input-multiple-output (MIMO) uplink with linear minimum mean-squared error (MMSE) channel estimation and detection. We utilize the angular domain channel representation for uniform linear antenna arrays, and adopt its equivalent independent and nonidentical distributed channel model. 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 achievable sum-rate. For tractable analysis and low-complexity solution, a tight approximation on the achievable sum-rate is derived first. Then the training length optimization for fixed training power and the training power optimization for fixed training length with respect to the approximate sum-rate maximization are both shown to be concave. An alternative optimization that solves the training length and power iteratively is proposed for the joint resource allocation. In addition, for the special case that the training and data transmission powers are equal, we derive the optimal training lengths for both high and low signal-to-noise- ratio (SNR) regions. Numerical results show the tightness of the derived sum-rate approximation and also the significant performance advantage of the proposed resource allocation.

摘要

创新点

采用等效于角度域信道模型的独立不同分布信道模型, 重点讨论了导频长度及功率联合分配。 针对大规模多输入多输出 (MIMO) 上行导频辅助传输, 得到线性最小均方误差接收可达和速率近似式, 无需迭代数值运算。 分别证明了固定导频长度下功率分配问题及固定导频功率下导频长度分配问题的凹函数性质, 进而提出针对联合分配问题的替换优化算法。 该算法具有计算复杂度低的特点。 此外, 还讨论了无功率分配传输下导频长度。 仿真结果证实了可达和速率表达的准确性及联合资源分配算法的有效性。