Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Study of Clustering Techniques and Scheduling Algorithms with Fairness for Cell-Free MIMO Networks

S. Mashdour, R. C. de Lamare

TL;DR

This work tackles AP clustering and downlink scheduling in cell-free MIMO networks by introducing information-rate-based clustering (Boosted SR, BSR) and a fairness-aware scheduling algorithm (F-Gr). APs for each UE are selected not only on large-scale fading but by the actual achievable per-link rates, with BSR augmenting coverage to meet a target average SR. A fairness-driven resource allocation problem is then solved with a Fair Greedy scheduling approach that alternates between high-rate selections and coverage-filling actions over time-slots. Simulation results show that BSR yields substantial sum-rate gains over LSF and that F-Gr achieves near CF performance with reduced complexity, highlighting practical gains for dense, multi-UE deployments.

Abstract

In this work, we propose a clustering technique based on information rates for cell-free massive multiple-input multiple-output (MIMO) networks. Unlike existing clustering approaches that rely on the large scale fading coefficients of the channels and user-centric techniques, we develop an approach that is based on the information rates of cell-free massive MIMO networks. We also devise a resource allocation technique to incorporate the proposed clustering and schedule users with fairness. An analysis of the proposed clustering approach based on information rates is carried out along with an assessment of its benefits for scheduling. Numerical results show that the proposed techniques outperform existing approaches.

Study of Clustering Techniques and Scheduling Algorithms with Fairness for Cell-Free MIMO Networks

TL;DR

This work tackles AP clustering and downlink scheduling in cell-free MIMO networks by introducing information-rate-based clustering (Boosted SR, BSR) and a fairness-aware scheduling algorithm (F-Gr). APs for each UE are selected not only on large-scale fading but by the actual achievable per-link rates, with BSR augmenting coverage to meet a target average SR. A fairness-driven resource allocation problem is then solved with a Fair Greedy scheduling approach that alternates between high-rate selections and coverage-filling actions over time-slots. Simulation results show that BSR yields substantial sum-rate gains over LSF and that F-Gr achieves near CF performance with reduced complexity, highlighting practical gains for dense, multi-UE deployments.

Abstract

In this work, we propose a clustering technique based on information rates for cell-free massive multiple-input multiple-output (MIMO) networks. Unlike existing clustering approaches that rely on the large scale fading coefficients of the channels and user-centric techniques, we develop an approach that is based on the information rates of cell-free massive MIMO networks. We also devise a resource allocation technique to incorporate the proposed clustering and schedule users with fairness. An analysis of the proposed clustering approach based on information rates is carried out along with an assessment of its benefits for scheduling. Numerical results show that the proposed techniques outperform existing approaches.
Paper Structure (13 sections, 19 equations, 4 figures, 1 table, 2 algorithms)

This paper contains 13 sections, 19 equations, 4 figures, 1 table, 2 algorithms.

Table of Contents

  1. Introduction
  2. System Models
  3. CF-mMIMO Network Model
  4. UCCF Network Model
  5. Proposed Clustering Based on Information Rates
  6. Clustering Based on Large Scale Fading
  7. Proposed Clustering Based on the Sum-Rate
  8. Analysis of the BSR Algorithm
  9. Resource Allocation with Fairness
  10. Proposed Resource Allocation Problem
  11. Proposed F-Gr Algorithm
  12. Simulation Results
  13. Conclusion

Figures (4)

  • Figure 1: User-centric cell free network.
  • Figure 2: Performance of networks: (a) Sum-rate for UCCF with different clustering criteria and CF network with no user scheduling, $L=16$, $N=4$, $K=128$, (b): BER of UCCF networks with different clustering criteria and the CF network with no user scheduling, $L=16$, $N=4$, $K=128$.
  • Figure 3: Performance and complexity of F-Gr resource allocation: (a) F-Gr resource allocation sum-rate in CF and UCCF networks, $L=16$, $N=4$, $K=128$ and $n=20$, (b) Complexity of the proposed resource allocation for CF and UCCF networks when $n=LN$ UEs are scheduled.
  • Figure 4: Scheduling times for each UE in CF network using Gr and F-Gr scheduling methods for $K=128$, $n=20$ and 100 channel realizations.