1: \begin{abstract}

2: In this paper, we consider the (global and sum) energy efficiency

3: optimization problem in downlink multi-input multi-output multi-cell

4: systems, where all users suffer from multi-user interference. This

5: is a challenging problem due to several reasons: 1) it is a nonconvex

6: fractional programming problem; 2) the transmission rate functions

7: are characterized by (complex-valued) transmit covariance matrices;

8: and 3) the processing-related power consumption may depend on the

9: transmission rate. We tackle this problem by the successive pseudoconvex

10: approximation approach, and we argue that pseudoconvex optimization

11: plays a fundamental role in designing novel iterative algorithms,

12: not only because every locally optimal point of a pseudoconvex optimization

13: problem is also globally optimal, but also because a descent direction

14: is easily obtained from every optimal point of a pseudoconvex optimization

15: problem. The proposed algorithms have the following advantages: 1)

16: fast convergence as the structure of the original optimization problem

17: is preserved as much as possible in the approximate problem solved

18: in each iteration, 2) easy implementation as each approximate problem

19: is suitable for parallel computation and its solution has a closed-form

20: expression, and 3) guaranteed convergence to a stationary point or

21: a Karush-Kuhn-Tucker point. The advantages of the proposed algorithm

22: are also illustrated numerically.

23: \end{abstract}