1: \begin{abstract}

2: %=======Motivation======================

3: 

4: % =======General Purpose of this paper==========

5: %We consider the problem of maximizing the sum rate of the MIMO-MAC with per-antenna power constraint. 

6: % =======How to solve this problem ======

7: %An efficient algorithm is proposed to compute optimal input covariance matrices and sum capacity.

8: We propose an iterative mode-dropping algorithm that optimizes input signals to achieve the sum capacity of the MIMO-MAC with per-antenna power constraint.

9: % =======Numerical issue==============

10: %The numerical algorithm is similar iterative water-filling for MIMO MAC with sum power constraint and iteratively maximizing the sum-rate objective by successively solving single-user MIMO optimization problem with per-antenna power constraint.

11: %In a way similar to iterative water-filling under sum power constraint, 

12: %It successively optimizes one of these matrices by

13: %applying mode-dropping to the equivalent single-user MIMO rate maximization problems.

14: %The algorithm converges to the sum capacity fast and precisely.

15: The algorithm successively optimizes each user's input covariance matrix by applying mode-dropping to the equivalent single-user MIMO rate maximization problem.

16: Both analysis and simulation show fast convergence. We then use the algorithm to briefly highlight the difference in MIMO-MAC capacities under sum and per-antenna power constraints.

17: %is based on the single-user MIMO rate optimization with per-antenna power constraint and 

18: %% =======Results: This turn out to be good==================

19: %The algorithm converges to the sum capacity fast and precisely.

20: %We also compare the capacity of MIMO-MAC under per-antenna power constraint with that under sum power constraint.

21: %and shows that with equal constraint per antenna, sum capacity can be close to the capacity with sum power constraint.

22: \end{abstract}

23: