Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Learning for Cross-Layer Resource Allocation in MEC-Aided Cell-Free Networks

Chong Zheng, Shiwen He, Yongming Huang, Tony Q. S. Quek

TL;DR

This work tackles cross-layer resource allocation in MEC-aided CF networks by jointly optimizing subcarrier allocation and beamforming to maximize the weighted sum rate $R_{\text{WSR}}=\sum_i \alpha_i r_i$. It introduces a multi-task learning formulation that recasts the optimization into $I+B+1$ self-supervised tasks and proposes centralized (CMTSSL) and distributed (DMTSSL) multi-task learners, guided by novel NFL and EL losses with adaptive uncertainty weights. A distance-aware transfer learning (DATL) mechanism is added to handle dynamic addition of SBSs with negligible cost, leveraging knowledge from the nearest existing SBS. Numerical results in a 3GPP 38.901 UMa scenario show that the proposed methods surpass baselines in sum-rate performance while offering scalable complexity and robustness to label noise and network dynamics. The approach provides a practical pathway to real-time, scalable cross-layer optimization in future MEC-aided wireless networks.

Abstract

Cross-layer resource allocation over mobile edge computing (MEC)-aided cell-free networks can sufficiently exploit the transmitting and computing resources to promote the data rate. However, the technical bottlenecks of traditional methods pose significant challenges to cross-layer optimization. In this paper, joint subcarrier allocation and beamforming optimization are investigated for the MEC-aided cell-free network from the perspective of deep learning to maximize the weighted sum rate. Specifically, we convert the underlying problem into a joint multi-task optimization problem and then propose a centralized multi-task self-supervised learning algorithm to solve the problem so as to avoid costly manual labeling. Therein, two novel and general loss functions, i.e., negative fraction linear loss and exponential linear loss whose advantages in robustness and target domain have been proved and discussed, are designed to enable self-supervised learning. Moreover, we further design a MEC-enabled distributed multi-task self-supervised learning (DMTSSL) algorithm, with low complexity and high scalability to address the challenge of dimensional disaster. Finally, we develop the distance-aware transfer learning algorithm based on the DMTSSL algorithm to handle the dynamic scenario with negligible computation cost. Simulation results under $3$rd generation partnership project 38.901 urban-macrocell scenario demonstrate the superiority of the proposed algorithms over the baseline algorithms.

Learning for Cross-Layer Resource Allocation in MEC-Aided Cell-Free Networks

TL;DR

This work tackles cross-layer resource allocation in MEC-aided CF networks by jointly optimizing subcarrier allocation and beamforming to maximize the weighted sum rate . It introduces a multi-task learning formulation that recasts the optimization into self-supervised tasks and proposes centralized (CMTSSL) and distributed (DMTSSL) multi-task learners, guided by novel NFL and EL losses with adaptive uncertainty weights. A distance-aware transfer learning (DATL) mechanism is added to handle dynamic addition of SBSs with negligible cost, leveraging knowledge from the nearest existing SBS. Numerical results in a 3GPP 38.901 UMa scenario show that the proposed methods surpass baselines in sum-rate performance while offering scalable complexity and robustness to label noise and network dynamics. The approach provides a practical pathway to real-time, scalable cross-layer optimization in future MEC-aided wireless networks.

Abstract

Cross-layer resource allocation over mobile edge computing (MEC)-aided cell-free networks can sufficiently exploit the transmitting and computing resources to promote the data rate. However, the technical bottlenecks of traditional methods pose significant challenges to cross-layer optimization. In this paper, joint subcarrier allocation and beamforming optimization are investigated for the MEC-aided cell-free network from the perspective of deep learning to maximize the weighted sum rate. Specifically, we convert the underlying problem into a joint multi-task optimization problem and then propose a centralized multi-task self-supervised learning algorithm to solve the problem so as to avoid costly manual labeling. Therein, two novel and general loss functions, i.e., negative fraction linear loss and exponential linear loss whose advantages in robustness and target domain have been proved and discussed, are designed to enable self-supervised learning. Moreover, we further design a MEC-enabled distributed multi-task self-supervised learning (DMTSSL) algorithm, with low complexity and high scalability to address the challenge of dimensional disaster. Finally, we develop the distance-aware transfer learning algorithm based on the DMTSSL algorithm to handle the dynamic scenario with negligible computation cost. Simulation results under rd generation partnership project 38.901 urban-macrocell scenario demonstrate the superiority of the proposed algorithms over the baseline algorithms.

Paper Structure

This paper contains 24 sections, 1 theorem, 31 equations, 9 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. SYSTEM MODEL AND PROBLEM FORMULATION
  3. System Model
  4. Problem Formulation
  5. Problem Recast
  6. METHODOLOGY
  7. Loss Design for Self-supervised Learning
  8. Centralized Self-supervised Multi-task Learning
  9. Loss schemes for the multiple tasks
  10. Training process of the CMTSSL algorithm
  11. Distributed Self-supervised Multi-task Learning
  12. Distance-aware Transfer Learning
  13. NUMERICAL SIMULATIONS AND ANALYSES
  14. Network Configurations
  15. Baselines
  16. ...and 9 more sections

Key Result

Theorem 1

In a general $Z$-class classification problem, the NFL loss and EL loss are both noice-tolerant to symmetric or uniform label noise when the noise rate $\eta<\frac{Z-1}{Z}$ as well as the loss input $x>x_1$ for the NFL loss or $x>x_2$ for the EL loss.

Figures (9)

  • Figure 1: Hierarchical architecture of the MEC-aided CF-MIMO system under investigation.
  • Figure 2: Illustration of the CMTSSL algorithm.
  • Figure 3: Illustration of the DMTSSL algorithm.
  • Figure 4: Illustration of the DATL algorithm.
  • Figure 5: Training losses of the proposed CMTSSL and DMTSSL algorithms on different loss schemes. ($B=3,I=10,N=4$)
  • ...and 4 more figures

Theorems & Definitions (1)

  • Theorem 1