Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Joint Sensing, Communication, and Computation for Vertical Federated Edge Learning in Edge Perception Network

Xiaowen Cao, Dingzhu Wen, Suzhi Bi, Yuanhao Cui, Guangxu Zhu, Han Hu, Yonina C. Eldar

TL;DR

The paper introduces an ISCC-enabled vertical federated edge learning framework for edge perception that leverages multi-view sensing data via feature-partitioned devices and over-the-air embeddings aggregation. It provides a theoretical convergence analysis under sensing and aggregation distortions and develops a practical alternating-optimization scheme to jointly tune batch size, sensing power, and transmission power, as well as the denoising factor. The model demonstrates faster convergence and higher recognition accuracy than baselines under realistic latency and energy constraints, illustrating the benefits of integrating sensing, communication, and computation in VFEEL. The work paves the way for efficient, privacy-preserving multi-view edge learning and inspires future extensions to decentralized consensus mechanisms.

Abstract

Combining wireless sensing and edge intelligence, edge perception networks enable intelligent data collection and processing at the network edge. However, traditional sample partition based horizontal federated edge learning struggles to effectively fuse complementary multiview information from distributed devices. To address this limitation, we propose a vertical federated edge learning (VFEEL) framework tailored for feature-partitioned sensing data. In this paper, we consider an integrated sensing, communication, and computation-enabled edge perception network, where multiple edge devices utilize wireless signals to sense environmental information for updating their local models, and the edge server aggregates feature embeddings via over-the-air computation for global model training. First, we analyze the convergence behavior of the ISCC-enabled VFEEL in terms of the loss function degradation in the presence of wireless sensing noise and aggregation distortions during AirComp.

Joint Sensing, Communication, and Computation for Vertical Federated Edge Learning in Edge Perception Network

TL;DR

The paper introduces an ISCC-enabled vertical federated edge learning framework for edge perception that leverages multi-view sensing data via feature-partitioned devices and over-the-air embeddings aggregation. It provides a theoretical convergence analysis under sensing and aggregation distortions and develops a practical alternating-optimization scheme to jointly tune batch size, sensing power, and transmission power, as well as the denoising factor. The model demonstrates faster convergence and higher recognition accuracy than baselines under realistic latency and energy constraints, illustrating the benefits of integrating sensing, communication, and computation in VFEEL. The work paves the way for efficient, privacy-preserving multi-view edge learning and inspires future extensions to decentralized consensus mechanisms.

Abstract

Combining wireless sensing and edge intelligence, edge perception networks enable intelligent data collection and processing at the network edge. However, traditional sample partition based horizontal federated edge learning struggles to effectively fuse complementary multiview information from distributed devices. To address this limitation, we propose a vertical federated edge learning (VFEEL) framework tailored for feature-partitioned sensing data. In this paper, we consider an integrated sensing, communication, and computation-enabled edge perception network, where multiple edge devices utilize wireless signals to sense environmental information for updating their local models, and the edge server aggregates feature embeddings via over-the-air computation for global model training. First, we analyze the convergence behavior of the ISCC-enabled VFEEL in terms of the loss function degradation in the presence of wireless sensing noise and aggregation distortions during AirComp.

Paper Structure

This paper contains 29 sections, 7 theorems, 61 equations, 8 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Contribution
  4. System Model
  5. Vertical Federated Edge Learning Algorithm
  6. Sensing Model for Training Samples Acquisition
  7. Local Computation for V-FEEL
  8. Embeddings Aggregation via Over-the-air Computation
  9. Network Resource Constraints
  10. Latency Constraints
  11. Transmission Power Constraints
  12. Energy Consumption Constraints
  13. Convergence Analysis of ISCC-based V-FEEL
  14. Basic Assumptions
  15. Data and Aggregation Error Analysis
  16. ...and 14 more sections

Key Result

Lemma 1

At each round $t$, for edge device $k\in\mathcal{K}$, the partial derivative of is unbiased: Then the variances of the partial derivatives are bounded as

Figures (8)

  • Figure 1: VFEEL versus HFEEL Yang19FL.
  • Figure 2: Illustration of ISCC-enabled VFEEL system.
  • Figure 3: Example local view of a global model in V-FEEL at each round.
  • Figure 4: Learning performance of ISCC-enabled VFEEL over different batch size, where the batch size is around 150 after optimization in the proposed scheme.
  • Figure 5: Learning performance of ISCC-enabled VFEEL over different channel noise variance.
  • ...and 3 more figures

Theorems & Definitions (12)

  • Remark 1
  • Lemma 1: Unbiased and Bounded Embedding Gradient Vector
  • Lemma 2: Per-Round Loss Function Reduction
  • Remark 2
  • Theorem 1: Convergence with Fixed Learning Rate
  • Remark 3
  • Proposition 1
  • Remark 4
  • Proposition 2
  • Lemma 3
  • ...and 2 more