Two-regimes interference classifier: An interference-aware resource allocation algorithm

Performance of heterogeneous network is strongly limited by the interference due to multiple access points operating in the same geographical area, with overlapping service coverage. The common understanding is that interference, classically processed as additive noise, compromises the transmission and therefore must be ideally avoided or at least strongly limited. However, recent investigations in the domain of information theory and successive interference cancellation (SIC) techniques have proved that interference may not necessarily be treated as an opponent, but may become an ally. In this paper, we propose a novel interference aware resource management algorithm, where the system may only control its interference perception. In a system consisting of a couple of downlink users and access points with overlapping coverage, we aim to define the most spectral-efficient way to process interference at each receiver. Based on a 3-regimes interference classifier, both users in the system may either treat interference as noise, orthogonalize transmissions so that interference may be avoided, or cancel interference out of the received signal via SIC-based techniques. Our study shows that, when aiming at maximizing total spectral efficiency, ignoring or avoiding interference is not always the best option. Based on our theoritical study, we propose an interference classification algorithm, with only 2 admissible regimes for each user. Finally, we assess its notable performance improvement by simulation results.