We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Fingerprinting Positioning in Distributed Massive MIMO Systems Using Affinity Propagation Clustering and Gaussian Process Regression.
- Authors
Moosavi, Seyedeh Samira; Fortier, Paul
- Abstract
Massive multiple-input multiple-output (M-MIMO) systems improve positioning accuracy besides enhancing communication performance. Fingerprinting (FP) method is widely used for positioning applications due to its high reliability, cost-efficiency, and accuracy. The FP method based on Gaussian process regression (GPR) could potentially be used in M-MIMO systems to improve positioning accuracy. However, it is limited by high computational complexity. In this paper, an FP positioning method based on the affinity propagation clustering (APC) and GPR is presented to estimate the user's position in a distributed massive MIMO (DM-MIMO) system from the uplink received signal strength (RSS). In the proposed method, an optimal clustering scheme based on APC is presented to split up the target area into several small regions, which minimizes the searching space of reference points and reduces the computational complexity and position estimation error. Then, a GPR model is created for each region based on the RSS data distribution within each region to provide further positioning accuracy. An improved method based on the K-dimensional tree (KD-tree) is also presented for test users to find their most likely region. Then their positions are estimated based on the GPR model of that region. Simulation results reveal that the proposed scheme improves positioning accuracy significantly compared to using only GPR for the whole target area. This approach has high coverage and improves average root-mean-squared error (RMSE) performance to a few meters, which is expected in 5G networks. Consequently, it also helps to reduce the computational complexity of GPR in the positioning systems.
- Subjects
KRIGING; MIMO systems; GAUSSIAN processes; COMPUTATIONAL complexity; DATA distribution; 5G networks
- Publication
Wireless Personal Communications, 2021, Vol 121, Issue 3, p1835
- ISSN
0929-6212
- Publication type
Article
- DOI
10.1007/s11277-021-08741-4