Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Joint Model Pruning and Resource Allocation for Wireless Time-triggered Federated Learning

Xinlu Zhang, Yansha Deng, Toktam Mahmoodi

TL;DR

This paper tackles scalability in wireless time-triggered federated learning by introducing TT-Prune, a joint framework that optimizes per-tier pruning ratios and wireless bandwidth to minimize the gradient norm under latency constraints. It provides a convergence bound for TT-Fed with adaptive pruning and derives closed-form, KKT-based solutions for pruning and bandwidth allocations, enabling efficient, dynamic adaptation. Empirical results on non-IID data with CNN models demonstrate about a 40% reduction in communication cost while maintaining convergence accuracy. The work offers significant practical impact for deploying TT-Fed in bandwidth-limited, large-scale wireless networks by balancing computation, communication, and learning performance.

Abstract

Time-triggered federated learning, in contrast to conventional event-based federated learning, organizes users into tiers based on fixed time intervals. However, this network still faces challenges due to a growing number of devices and limited wireless bandwidth, increasing issues like stragglers and communication overhead. In this paper, we apply model pruning to wireless Time-triggered systems and jointly study the problem of optimizing the pruning ratio and bandwidth allocation to minimize training loss under communication latency constraints. To solve this joint optimization problem, we perform a convergence analysis on the gradient $l_2$-norm of the asynchronous multi-tier federated learning (FL) model with adaptive model pruning. The convergence upper bound is derived and a joint optimization problem of pruning ratio and wireless bandwidth is defined to minimize the model training loss under a given communication latency constraint. The closed-form solutions for wireless bandwidth and pruning ratio by using KKT conditions are then formulated. As indicated in the simulation experiments, our proposed TT-Prune demonstrates a 40% reduction in communication cost, compared with the asynchronous multi-tier FL without model pruning, while maintaining the model convergence at the same level.

Joint Model Pruning and Resource Allocation for Wireless Time-triggered Federated Learning

TL;DR

This paper tackles scalability in wireless time-triggered federated learning by introducing TT-Prune, a joint framework that optimizes per-tier pruning ratios and wireless bandwidth to minimize the gradient norm under latency constraints. It provides a convergence bound for TT-Fed with adaptive pruning and derives closed-form, KKT-based solutions for pruning and bandwidth allocations, enabling efficient, dynamic adaptation. Empirical results on non-IID data with CNN models demonstrate about a 40% reduction in communication cost while maintaining convergence accuracy. The work offers significant practical impact for deploying TT-Fed in bandwidth-limited, large-scale wireless networks by balancing computation, communication, and learning performance.

Abstract

Time-triggered federated learning, in contrast to conventional event-based federated learning, organizes users into tiers based on fixed time intervals. However, this network still faces challenges due to a growing number of devices and limited wireless bandwidth, increasing issues like stragglers and communication overhead. In this paper, we apply model pruning to wireless Time-triggered systems and jointly study the problem of optimizing the pruning ratio and bandwidth allocation to minimize training loss under communication latency constraints. To solve this joint optimization problem, we perform a convergence analysis on the gradient -norm of the asynchronous multi-tier federated learning (FL) model with adaptive model pruning. The convergence upper bound is derived and a joint optimization problem of pruning ratio and wireless bandwidth is defined to minimize the model training loss under a given communication latency constraint. The closed-form solutions for wireless bandwidth and pruning ratio by using KKT conditions are then formulated. As indicated in the simulation experiments, our proposed TT-Prune demonstrates a 40% reduction in communication cost, compared with the asynchronous multi-tier FL without model pruning, while maintaining the model convergence at the same level.
Paper Structure (18 sections, 22 equations, 3 figures)

This paper contains 18 sections, 22 equations, 3 figures.

Table of Contents

  1. Introduction
  2. System Model and problem formulation
  3. Asynchronous multi-tier FL Framework
  4. Global Model Broadcast
  5. Local Model Learning
  6. Intra-tier Aggregation
  7. Global Aggregation
  8. Federated Learning With Model Pruning
  9. Latency In Each Global Iteration
  10. Local Model Update With Pruning
  11. Local Model Uplink Transmission
  12. Global Model Aggregation and Broadcast
  13. Problem Formulation
  14. Convergence Analysis and Optimal Solution
  15. Convergence analysis
  16. ...and 3 more sections

Figures (3)

  • Figure 1: The work-flow of Pruned TT-Fed under the given aggregation duration $\Delta T$.
  • Figure 2: Test accuracy of TT-Fed under different scheme in Non-IID FMNIST dataset
  • Figure 3: Performance required for TT-Prune and other schemes to achieve 80% accuracy in Non-IID FMNIST dataset