K-Means Pilot Assignment for Scalable Cell-Free Massive MIMO

Cell-free Massive MIMO is a promising technology for enhancing wireless network performance. However, it suffers from pilot contamination, where multiple users interfere with each other due to the reuse of identical identification signals. This paper introduces a clustering approach to mitigate pilot contamination by jointly optimizing user clustering and pilot assignment. We employ the K-means clustering algorithm to group users based on their geographical locations and channel conditions, thereby minimizing interference within clusters. By allocating unique pilot sequences to each user within a cluster, we effectively reduce pilot contamination. This approach works in tandem with access point clustering. Initially, the channel gain between each access point and each user is calculated. These channel gain values are then summed up to determine the overall communication efficiency. Access points are chosen to be part of the cluster only if their individual channel gain is 5% or more of the total aggregated channel gain. This ensures that only APs with substantial channel contributions are involved in serving the UEs. Our results demonstrate that this integrated approach significantly improves spectral efficiency compared to conventional methods, leading to higher data rates and better overall network performance.