# RIFLE: Robust Distillation-based FL for Deep Model Deployment on Resource-Constrained IoT Networks

Source: [RIFLE: Robust Distillation-based FL for Deep Model Deployment on Resource-Constrained IoT Networks](<https://www.sota2.com/research/paper/698abbe1c97a0b3e3b5de2a4>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-11T05:00:13.386Z

Authors: Pouria Arefijamal, Mahdi Ahmadlou, Bardia Safaei, J\\"org Henkel

Published: 2026-02-09T09:57:59.000Z

## Abstract

Federated learning (FL) is a decentralized learning paradigm widely adopted in resource-constrained Internet of Things (IoT) environments. These devices, typically relying on TinyML models, collaboratively train global models by sharing gradients with a central server while preserving data privacy. However, as data heterogeneity and task complexity increase, TinyML models often become insufficient to capture intricate patterns, especially under extreme non-IID (non-independent and identically distributed) conditions. Moreover, ensuring robustness against malicious clients and poisoned updates remains a major challenge. Accordingly, this paper introduces RIFLE - a Robust, distillation-based Federated Learning framework that replaces gradient sharing with logit-based knowledge transfer. By leveraging a knowledge distillation aggregation scheme, RIFLE enables the training of deep models such as VGG-19 and Resnet18 within constrained IoT systems. Furthermore, a Kullback-Leibler (KL) divergence-based validation mechanism quantifies the reliability of client updates without exposing raw data, achieving high trust and privacy preservation simultaneously. Experiments on three benchmark datasets (MNIST, CIFAR-10, and CIFAR-100) under heterogeneous non-IID conditions demonstrate that RIFLE reduces false-positive detections by up to 87.5%, enhances poisoning attack mitigation by 62.5%, and achieves up to 28.3% higher accuracy compared to conventional federated learning baselines within only 10 rounds. Notably, RIFLE reduces VGG19 training time from over 600 days to just 1.39 hours on typical IoT devices (0.3 GFLOPS), making deep learning practical in resource-constrained networks.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Image Classification on CIFAR-100 (test) (Robustness Metrics)](<https://www.sota2.com/research/sota/image-classification-on-cifar-100-test-robustness-metrics.md>): Image Classification; CIFAR-100 (test); ASR: 19.

- [Image Classification on MNIST (test) (ASR and Robustness Metrics)](<https://www.sota2.com/research/sota/image-classification-on-mnist-test-asr-and-robustness-metrics.md>): Image Classification; MNIST (test); ASR: 12.

- [Image Classification on CIFAR-10 (test) (Robustness Metrics)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-test-robustness-metrics.md>): Image Classification; CIFAR-10 (test); ASR: 0.16.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2602.08446>)

[PDF](<https://arxiv.org/pdf/2602.08446.pdf>)
