# FLEX-SFL: A Flexible and Efficient Split Federated Learning Framework for Edge Heterogeneity

## TL;DR

FLEX-SFL is a new framework for edge computing that improves training efficiency and scalability in heterogeneous environments using adaptive model segmentation and client selection.

## Contribution

FLEX-SFL introduces dynamic, device-aware model segmentation and entropy-driven client selection to enhance federated learning in edge environments.

## Key findings

- FLEX-SFL outperforms existing methods in accuracy, convergence speed, and resource efficiency on multiple datasets.
- The framework's mechanisms improve training throughput and reduce communication latency in heterogeneous edge environments.

## Abstract

What are the main findings?
The FLEX-SFL framework introduces dynamic, device-aware adaptive model segmentation, entropy-driven client selection, and hierarchical local asynchronous aggregation mechanisms, improving training efficiency and scalability in edge heterogeneous environments.Extensive experiments demonstrate that FLEX-SFL outperforms state-of-the-art federated and split federated learning methods in terms of accuracy, convergence speed, and resource efficiency across multiple datasets.

The FLEX-SFL framework introduces dynamic, device-aware adaptive model segmentation, entropy-driven client selection, and hierarchical local asynchronous aggregation mechanisms, improving training efficiency and scalability in edge heterogeneous environments.

Extensive experiments demonstrate that FLEX-SFL outperforms state-of-the-art federated and split federated learning methods in terms of accuracy, convergence speed, and resource efficiency across multiple datasets.

What are the implications of the main findings?
FLEX-SFL provides a practical solution to the challenges posed by system and statistical heterogeneity in federated learning, making it suitable for large-scale edge deployments in real-world intelligent systems.The proposed mechanisms can be extended to enhance the scalability and adaptability of other federated learning frameworks, potentially improving edge computing applications in fields like IoT and healthcare.

FLEX-SFL provides a practical solution to the challenges posed by system and statistical heterogeneity in federated learning, making it suitable for large-scale edge deployments in real-world intelligent systems.

The proposed mechanisms can be extended to enhance the scalability and adaptability of other federated learning frameworks, potentially improving edge computing applications in fields like IoT and healthcare.

The deployment of Federated Learning (FL) in edge environments is often impeded by system heterogeneity, non-independent and identically distributed (non-IID) data, and constrained communication resources, which collectively hinder training efficiency and scalability. To address these challenges, this paper presents FLEX-SFL, a flexible and efficient split federated learning framework that jointly optimizes model partitioning, client selection, and communication scheduling. FLEX-SFL incorporates three coordinated mechanisms: a device-aware adaptive segmentation strategy that dynamically adjusts model partition points based on client computational capacity to mitigate straggler effects; an entropy-driven client selection algorithm that promotes data representativeness by leveraging label distribution entropy; and a hierarchical local asynchronous aggregation scheme that enables asynchronous intra-cluster and inter-cluster model updates to improve training throughput and reduce communication latency. We theoretically establish the convergence properties of FLEX-SFL under convex settings and analyze the influence of local update frequency and client participation on convergence bounds. Extensive experiments on benchmark datasets including FMNIST, CIFAR-10, and CIFAR-100 demonstrate that FLEX-SFL consistently outperforms state-of-the-art FL and split FL baselines in terms of model accuracy, convergence speed, and resource efficiency, particularly under high degrees of statistical and system heterogeneity. These results validate the effectiveness and practicality of FLEX-SFL for real-world edge intelligent systems.