Abstract: A distributed power allocation scheme was presented to maximize the system capacity in dense small cell networks. A new signaling called inter-cell-signal to interference plus noise ratio (ISINR) as well as its modification was defined to show the algebraic properties of the system capacity. With the help of ISINR, we have an easy way to identify the local monotonicity of the system capacity. Then on each subchannel in iteration, we divide the small cell evolved node B’s (SeNBs) into different subsets. For the first subset, the sum rate is convex with respect to the power domain and the power optimally was allocated. On the other hand, for the second subset, the sum rate is monotone decreasing and the SeNBs would abandon the subchannel in this iteration. The two strategies are applied iteratively to improve the system capacity. Simulations show that the proposed scheme can achieve much larger system capacity than the conventional ones. The scheme can achieve a promising tradeoff between performance and signaling overhead.

Key words: power allocation | dense small cells |Inter-cell-signal to interference plus noise ratio |orthogonal frequency division multiple access