Skip to content

An Efficient Architecture for Edge AI Federated Learning With Homomorphic Encryption

Authors: Dadmehr Rahbari; Masoud Daneshtalab; Maksim Jenihhin
Year: 2025
Venue: IEEE Access

type: system collection: ppml status: auto

Links: DOI

Summary

Auto-extracted summary (unreviewed)

Proposes a federated-learning architecture for edge AI in which homomorphic encryption protects model updates during decentralized training, combined with dynamic client clustering, quantization-aware training and structured model pruning to cut communication, memory and convergence cost. On the Human Activity Recognition dataset the authors report an average +8.4% accuracy, 16.2% lower latency, 35.1% lower memory use and 2.7% lower security overhead than several state-of-the-art FL methods.

Abstract

With the rapid growth of edge AI applications, there is an increasing demand for federated learning (FL) frameworks that are both efficient and privacy-preserving. This work introduces a robust approach that leverages homomorphic encryption (HE) to ensure data confidentiality during decentralized training. To tackle the typical challenges of FL—such as high communication overhead, resource limitations, and convergence inefficiencies—our method integrates dynamic client clustering, quantization-aware training, and structured model pruning. These optimizations collectively reduce latency and memory consumption while accelerating model convergence. Evaluations using the Human Activity Recognition dataset show that the proposed approach outperforms several state-of-the-art FL methods, achieving an average increase of +8.4% in accuracy, −16.2% lower latency, −35.1% reduction in memory usage, and −2.7% lower security overhead. These results demonstrate its suitability for real-time, resource-constrained scenarios in domains like healthcare, IoT, and finance, where maintaining a strong balance between efficiency and privacy is essential.
Source: pdf

Extracted details

Field Value Source Confidence
contribution HE-protected federated learning framework for resource-constrained edge devices llm 0.60
contribution dynamic client clustering to reduce communication rounds llm 0.60
contribution quantization-aware training to reduce update size llm 0.60
contribution structured model pruning to reduce memory and latency llm 0.60
contribution evaluation against state-of-the-art FL methods on HAR data llm 0.60
schemes homomorphic encryption (scheme not named in excerpt) llm 0.60
datasets Human Activity Recognition (HAR) llm 0.60
tasks federated learning llm 0.60
tasks human activity recognition llm 0.60
tasks edge AI llm 0.60
interaction other llm 0.60
key result accuracy improvement: +8.4% (average) (HAR dataset vs state-of-the-art FL methods) llm 0.60
key result latency reduction: -16.2% (HAR dataset vs state-of-the-art FL methods) llm 0.60
key result memory usage reduction: -35.1% (HAR dataset vs state-of-the-art FL methods) llm 0.60
key result security overhead reduction: -2.7% (HAR dataset vs state-of-the-art FL methods) llm 0.60

Keyword mentions

Counts of vocabulary terms in the full text (first page of occurrence in brackets).

  • scheme: CKKS 7 [p.4], MPC 2 [p.1], Functional encryption 2 [p.2], Differential privacy 2 [p.1]
  • model: Logistic regression 1 [p.2], Decision tree 1 [p.10], GAN 1 [p.10]
  • dataset: MNIST 1 [p.2], CIFAR-10 1 [p.2]
  • technique: Quantization 32 [p.1], Federated learning 27 [p.1], Pruning 23 [p.1], Softmax 6 [p.3], ReLU 6 [p.3]
  • task: Clustering 13 [p.1], IoT / edge 12 [p.1], Medical / healthcare 8 [p.1], Finance 2 [p.1]
  • topic: Homomorphic encryption 42 [p.1]

Files

Path (under papers/) Collection Pages Size SHA-256 Note
references/ppml/federated-learning/rahbari-2025_efficient-architecture-for-edge-ai-federated-learning-with-he.pdf ppml/federated-learning 11 5.3 MB 20f64bd3d48c was Rahbari et al. - 2025 - An Efficient Architecture for Edge AI Federated Learning With Homomorphic Encryption.pdf

Database key: an-efficient-architecture-for-edge-ai-federated-learning-with · created 2026-11-05 · updated 2026-11-05 · edit with python3 tools/paperdb.py set an-efficient-architecture-for-edge-ai-federated-learning-with COLUMN VALUE