Power Control and Interference Management in Dense Wireless Networks

TL;DR

This paper introduces a fast, low-complexity fixed-point iteration algorithm for power control and interference management in dense wireless networks, applicable to various utility functions and scalable for distributed implementation.

Contribution

The paper proposes a novel fixed-point based power control algorithm with proven local convergence, reduced computational complexity, and suitability for distributed deployment in dense wireless networks.

Findings

Algorithm converges in less than 15 iterations

Computational complexity per iteration is O(N^2)

Distributed implementation achieves O(N) complexity per node

Abstract

We address the problem of interference management and power control in terms of maximization of a general utility function. For the utility functions under consideration, we propose a power control algorithm based on a fixed-point iteration; further, we prove local convergence of the algorithm in the neighborhood of the optimal power vector. Our algorithm has several benefits over the previously studied works in the literature: first, the algorithm can be applied to problems other than network utility maximization (NUM), e.g., power control in a relay network; second, for a network with $N$ wireless transmitters, the computational complexity of the proposed algorithm is $\mathcal{O}(N^2)$ calculations per iteration (significantly smaller than the $\mathcal{O}(N^3) $ calculations for Newton's iterations or gradient descent approaches). Furthermore, the algorithm converges very fast (usually in less than 15 iterations), and in particular, if initialized close to the optimal solution, the convergence speed is much faster. This suggests the potential of tracking variations in slowly fading channels. Finally, when implemented in a distributed fashion, the algorithm attains the optimal power vector with a signaling/computational complexity of only $\mathcal{O}(N)$ at each node.