Pilot Contamination Reduction Using Weighted Graph Coloring and Soft Pilot Reuse
Subject Areas : electrical and computer engineeringAmir Rostaei 1 , Hossein Khaleghi 2
1 - Emam Hossein University
2 -
Keywords: Pilot contamination, massive MIMO channel estimation, weighted graph coloring (WGC), edge-weighted interference graph (EWIG).,
Abstract :
Massive MIMO cellular networks, despite their ability to serve multiple users simultaneously, face a significant challenge due to pilot contamination. This paper presents an innovative two-stage algorithm to reduce this contamination and increase user data rates in both uplink and downlink. The key innovation of the proposed method lies in the intelligent integration of three techniques: Soft Pilot Reuse (SPR), optimal pilot sequence selection algorithm, and Weighted Graph Coloring (WGC). This combination simultaneously addresses three fundamental issues: contamination in SPR due to fixed thresholds, increased pilot overhead in WGC, and the problem of maximizing the data rate of the user with the lowest rate. The proposed algorithm operates in two stages. In the first stage, users are divided into center and edge groups, and optimal pilot sequences are determined based on their data rates. In the second stage, using the WGC algorithm and creating an Edge-Weighted Interference Graph (EWIG), pilot contamination is reduced based on the intensity of user interference. Simulation results show that this method significantly improves system performance compared to the best existing method (WGC). In the downlink, an 11 dB improvement in Signal-to-Interference-plus-Noise Ratio (SINR) and a 0.16 bps/Hz increase in average achievable rate are observed. In the uplink, a 2.4 dB improvement in SINR and a 0.46 bps/Hz increase in average achievable rate are achieved. In terms of computational complexity, the proposed method has lower complexity compared to the WGC scheme. Moreover, energy efficiency analysis confirms the superiority of the proposed method in both uplink and downlink scenarios.