Efficient Split Learning LSTM Models for FPGA-based Edge IoT Devices

2025-02-12

AI Summary

To address the performance–resource–energy trade-off in deploying time-series models on resource-constrained edge IoT devices—particularly FPGA platforms—this paper presents the first systematic co-design of LSTM and Split Learning (SL) for FPGA-based edge execution. We propose a hierarchical configuration methodology that jointly optimizes inference latency, power consumption, and hardware resource utilization, integrating model pruning, quantization, and edge-cloud collaborative training. Evaluated on a real FPGA edge platform, our implementation achieves sub-12-ms per-sample inference latency, under 1.8 W power consumption, and reduces RMSE by 23% over baselines in a water quality monitoring task—outperforming both end-to-end and conventional SL approaches. This work establishes a reproducible, lightweight, low-power, and high-accuracy deployment paradigm for time-series modeling at the edge via SL-FPGA co-design.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Natural Language Processing: Learning & Optimization for NLP

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Split Learning (SL) recently emerged as an efficient paradigm for distributed Machine Learning (ML) suitable for the Internet Of Things (IoT)-Cloud systems. However, deploying SL on resource-constrained edge IoT platforms poses a significant challenge in terms of balancing the model performance against the processing, memory, and energy resources. In this work, we present a practical study of deploying SL framework on a real-world Field-Programmable Gate Array (FPGA)-based edge IoT platform. We address the SL framework applied to a time-series processing model based on Recurrent Neural Networks (RNNs). Set in the context of river water quality monitoring and using real-world data, we train, optimize, and deploy a Long Short-Term Memory (LSTM) model on a given edge IoT FPGA platform in different SL configurations. Our results demonstrate the importance of aligning design choices with specific application requirements, whether it is maximizing speed, minimizing power, or optimizing for resource constraints.

Problem

• Deploying Split Learning on edge IoT devices
• Balancing model performance with resource constraints
• Optimizing LSTM models for FPGA-based systems

Innovation

• Split Learning for IoT
• LSTM on FPGA
• Resource-constrained optimization