Treating Interference as Noise in Cellular Networks: A Stochastic Geometry Approach

• Published in: IEEE transactions on wireless communications 2020-03, Vol.19 (3), p.1918-1932
• Main Authors: Bacha, Mudasar, Di Renzo, Marco, Clerckx, Bruno
• Format: Article
• Language: English
• Subjects: Algorithms; Cellular communication; Cellular networks; Engineering Sciences; Geometry; Interference; Noise; Optimization; Parameters; Poisson point processes; Probability theory; Scheduling; Scheduling algorithms; Signal and Image processing; Signal to noise ratio; stochastic geometry; system level analysis; Tin; Treating interference as noise (TIN); user scheduling
• ISSN: 1536-1276; 1558-2248
• DOI: 10.1109/TWC.2019.2959773

The interference management technique that treats interference as noise (TIN) is optimal when the interference is sufficiently weak. Scheduling algorithms based on the TIN optimality condition have recently been proposed, e.g., for application to device-to-device communications. TIN, however, has never been applied to cellular networks. In this work, we propose a scheduling algorithm for application to cellular networks that is based on the TIN optimality condition. In the proposed scheduling algorithm, each base station (BS) first randomly selects a user equipment (UE) in its coverage region, and then checks the TIN optimality conditions. If the latter conditions are not fulfilled, the BS is turned off. In order to assess the performance of TIN applied to cellular networks, we introduce an analytical framework with the aid of stochastic geometry theory. We develop, in particular, tractable expressions of the signal-to-interference-and-noise ratio (SINR) coverage probability and average rate of cellular networks. In addition, we carry out asymptotic analysis to find the optimal system parameters that maximize the SINR coverage probability. By using the optimized system parameters, it is shown that TIN applied to cellular networks yields significant gains in terms of SINR coverage probability and average rate. Specifically, the numerical results show that average rate gains of the order of 21% over conventional scheduling algorithms are obtained.