LEO-Split: A Semi-Supervised Split Learning Framework over LEO Satellite Networks

TL;DR

LEO-Split introduces a semi-supervised split learning framework designed for LEO satellite networks, addressing intermittent connectivity, data scarcity, and heterogeneity to enable efficient distributed deep learning.

Contribution

The paper proposes a novel semi-supervised split learning framework tailored for LEO satellite networks, incorporating auxiliary models, pseudo-labeling, and adaptive schemes to improve training under connectivity and data challenges.

Findings

Outperforms state-of-the-art benchmarks in real-world satellite traces

Effectively handles data scarcity and heterogeneity in satellite networks

Reduces overfitting through adaptive activation interpolation

Abstract

Recently, the increasing deployment of LEO satellite systems has enabled various space analytics (e.g., crop and climate monitoring), which heavily relies on the advancements in deep learning (DL). However, the intermittent connectivity between LEO satellites and ground station (GS) significantly hinders the timely transmission of raw data to GS for centralized learning, while the scaled-up DL models hamper distributed learning on resource-constrained LEO satellites. Though split learning (SL) can be a potential solution to these problems by partitioning a model and offloading primary training workload to GS, the labor-intensive labeling process remains an obstacle, with intermittent connectivity and data heterogeneity being other challenges. In this paper, we propose LEO-Split, a semi-supervised (SS) SL design tailored for satellite networks to combat these challenges. Leveraging SS learning to handle (labeled) data scarcity, we construct an auxiliary model to tackle the training failure of the satellite-GS non-contact time. Moreover, we propose a pseudo-labeling algorithm to rectify data imbalances across satellites. Lastly, an adaptive activation interpolation scheme is devised to prevent the overfitting of server-side sub-model training at GS. Extensive experiments with real-world LEO satellite traces (e.g., Starlink) demonstrate that our LEO-Split framework achieves superior performance compared to state-ofthe-art benchmarks.