Improved Sum Power Iterative Water-Filling with Rapid Convergence and Robustness for Multi-Antenna Gaussian Broadcast Channels

In our previous works, we have pointed out that when the iterative water-filling algorithms are applied to find the sum capacity of the multi-input multi-output broadcast channel (MIMO BC), it has some limitations. One of the major limitations is that the greater the number of the users becomes, the slower convergence of the iterative water-filling algorithms appear to be. To address this limitation, this paper first presents a new iterative water-filling algorithm for the MIMO BC. When compared with previous research, as the first advantage, the performance of this proposed algorithm has a strong robustness for the number of the users $K$; as the second advantage, parallel processing can be utilized to benefit the speed of computation during the implementation of this proposed algorithm. In addition, the proposed new algorithm is effective for finding the optimal transmission policy due to its simplicity and fast convergence.