Nonlinear Distortion-Aware Interference Alignment for Ultra-Dense Networks

Ultra-dense networks (UDNs) have been proposed to achieve high data rates and energy efficiency, but their performance is limited by the increase of inter-user interference. To overcome this problem, interference alignment (IA) algorithms have been widely researched. However, reported IA techniques neglect the nonlinear distortion (NLD) induced by power amplifiers. Thus, their performance is severely degraded in power-efficient transmissions operating close to the saturation point. Distortion-aware precoding techniques have been studied to reduce NLD in single transmitter scenarios either single-user or multi-user. Nevertheless, its extension to UDNs with multiple transmitters and users is not straightforward. In this work, a novel IA algorithm, named NLD-IA, is proposed to alternatively design the precoding and combining vectors to reduce the interference and NLD by maximizing the sum-rate through the signal-to-interference-plus-noise ratio (SINR). Precoding vectors are obtained via a non-convex optimization problem that models the NLD correlation. An analytical solution based on the Newton method over the complex field is developed to solve this problem with low computational complexity. Simulation results show that the proposed NLD-IA significantly reduces interference and NLD outperforming previous IA algorithms. The proposed method is an attractive solution for UDNs commonly found in the context of Internet of Things applications.