Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Split Federated Learning for UAV-Enabled Integrated Sensing, Computation, and Communication

Xiangwang Hou, Jingjing Wang, Zekai Zhang, Jiacheng Wang, Lei Liu, Yong Ren

TL;DR

This work tackles energy-efficient distributed learning for UAV-enabled ISCC by introducing Split Federated Learning for UAV-Enabled ISCC (SFLSCC), which partitions training between UAVs (client-side) and an edge server (server-side) to reduce UAV computation and privacy risk. The authors derive closed-form upper bounds on the convergence gap as functions of UAV deployment, split point $L_c$, sensing volume $b$, and aggregation frequency $I$, and formulate a joint energy-minimization problem to achieve a target accuracy. A low-complexity algorithm based on Newton-Raphson, traversing for $L_c$, and SCA/BCD is proposed to obtain near-optimal configurations. Extensive simulations on a target-motion recognition task demonstrate superior energy efficiency and competitive convergence relative to baselines, validating the practical impact for energy-constrained UAV FEL systems.

Abstract

Unmanned aerial vehicles (UAVs) with integrated sensing, computation, and communication (ISCC) capabilities have become key enablers of next-generation wireless networks. Federated edge learning (FEL) leverages UAVs as mobile learning agents to collect data, perform local model updates, and contribute to global model aggregation. However, existing UAV-assisted FEL systems face critical challenges, including excessive computational demands, privacy risks, and inefficient communication, primarily due to the requirement for full-model training on resource-constrained UAVs. To deal with aforementioned challenges, we propose Split Federated Learning for UAV-Enabled ISCC (SFLSCC), a novel framework that integrates split federated learning (SFL) into UAV-assisted FEL. SFLSCC optimally partitions model training between UAVs and edge servers, significantly reducing UAVs' computational burden while preserving data privacy. We conduct a theoretical analysis of UAV deployment, split point selection, data sensing volume, and client-side aggregation frequency, deriving closed-form upper bounds for the convergence gap. Based on these insights, we conceive a joint optimization problem to minimize the energy consumption required to achieve a target model accuracy. Given the non-convex nature of the problem, we develop a low-complexity algorithm to efficiently determine UAV deployment, split point selection, and communication frequency. Extensive simulations on a target motion recognition task validate the effectiveness of SFLSCC, demonstrating superior convergence performance and energy efficiency compared to baseline methods.

Split Federated Learning for UAV-Enabled Integrated Sensing, Computation, and Communication

TL;DR

This work tackles energy-efficient distributed learning for UAV-enabled ISCC by introducing Split Federated Learning for UAV-Enabled ISCC (SFLSCC), which partitions training between UAVs (client-side) and an edge server (server-side) to reduce UAV computation and privacy risk. The authors derive closed-form upper bounds on the convergence gap as functions of UAV deployment, split point , sensing volume , and aggregation frequency , and formulate a joint energy-minimization problem to achieve a target accuracy. A low-complexity algorithm based on Newton-Raphson, traversing for , and SCA/BCD is proposed to obtain near-optimal configurations. Extensive simulations on a target-motion recognition task demonstrate superior energy efficiency and competitive convergence relative to baselines, validating the practical impact for energy-constrained UAV FEL systems.

Abstract

Unmanned aerial vehicles (UAVs) with integrated sensing, computation, and communication (ISCC) capabilities have become key enablers of next-generation wireless networks. Federated edge learning (FEL) leverages UAVs as mobile learning agents to collect data, perform local model updates, and contribute to global model aggregation. However, existing UAV-assisted FEL systems face critical challenges, including excessive computational demands, privacy risks, and inefficient communication, primarily due to the requirement for full-model training on resource-constrained UAVs. To deal with aforementioned challenges, we propose Split Federated Learning for UAV-Enabled ISCC (SFLSCC), a novel framework that integrates split federated learning (SFL) into UAV-assisted FEL. SFLSCC optimally partitions model training between UAVs and edge servers, significantly reducing UAVs' computational burden while preserving data privacy. We conduct a theoretical analysis of UAV deployment, split point selection, data sensing volume, and client-side aggregation frequency, deriving closed-form upper bounds for the convergence gap. Based on these insights, we conceive a joint optimization problem to minimize the energy consumption required to achieve a target model accuracy. Given the non-convex nature of the problem, we develop a low-complexity algorithm to efficiently determine UAV deployment, split point selection, and communication frequency. Extensive simulations on a target motion recognition task validate the effectiveness of SFLSCC, demonstrating superior convergence performance and energy efficiency compared to baseline methods.

Paper Structure

This paper contains 27 sections, 6 theorems, 73 equations, 6 figures.

Table of Contents

  1. Introduction
  2. Related Works
  3. Contributions
  4. System Model
  5. Split Federated Learning Model
  6. Sensed data
  7. Forward propagation of client-side
  8. Forward and backward propagation of server-side
  9. Client-side model updating
  10. Aggregation of client-side
  11. UAV Communication Model
  12. Target Sensing Model
  13. Energy Consumption Model
  14. Energy consumption of target sensing
  15. Energy consumption of local training
  16. ...and 12 more sections

Key Result

Lemma 1

According to Assumption 1, we can derive

Figures (6)

  • Figure 1: The architecture of SFLSCC.
  • Figure 2: Spectrograms of different motions generated from simulator9593198.
  • Figure 3: Comparison of energy consumption and convergence performance across different schemes.
  • Figure 4: The impact of different optimization variables on energy consumption of SFLSCC.
  • Figure 5: The impact of different optimization variables on the convergence of SFLSCC.
  • ...and 1 more figures

Theorems & Definitions (12)

  • Lemma 1
  • proof
  • Lemma 2
  • proof
  • Lemma 3
  • proof
  • Theorem 1
  • proof
  • Corollary 1
  • proof
  • ...and 2 more