Automated Energy-Aware Time-Series Model Deployment on Embedded FPGAs for Resilient Combined Sewer Overflow Management

2025-08-19

AI Summary

To address the insufficient reliability of combined sewer overflow (CSO) prediction under climate change, this paper proposes an edge-intelligence-oriented, end-to-end, low-power forecasting framework. Methodologically, we design a hardware-aware automated deployment pipeline that jointly optimizes inference accuracy and energy efficiency, achieving, for the first time on resource-constrained AMD Spartan-7 FPGAs, fully integer-quantized inference for both Transformer and LSTM models. Key contributions include: (1) overcoming edge-device computational bottlenecks—8-bit quantized Transformer achieves MSE 0.0376 on real-world data with only 0.370 mJ per inference; (2) ultra-low-energy LSTM inference at 0.009 mJ—40× more efficient than the Transformer—enabling long-term, battery-operated deployment; and (3) localized real-time CSO alerts under extreme conditions (e.g., communication outages), significantly enhancing system resilience.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Extreme weather events, intensified by climate change, increasingly challenge aging combined sewer systems, raising the risk of untreated wastewater overflow. Accurate forecasting of sewer overflow basin filling levels can provide actionable insights for early intervention, helping mitigating uncontrolled discharge. In recent years, AI-based forecasting methods have offered scalable alternatives to traditional physics-based models, but their reliance on cloud computing limits their reliability during communication outages. To address this, we propose an end-to-end forecasting framework that enables energy-efficient inference directly on edge devices. Our solution integrates lightweight Transformer and Long Short-Term Memory (LSTM) models, compressed via integer-only quantization for efficient on-device execution. Moreover, an automated hardware-aware deployment pipeline is used to search for optimal model configurations by jointly minimizing prediction error and energy consumption on an AMD Spartan-7 XC7S15 FPGA. Evaluated on real-world sewer data, the selected 8-bit Transformer model, trained on 24 hours of historical measurements, achieves high accuracy (MSE 0.0376) at an energy cost of 0.370 mJ per inference. In contrast, the optimal 8-bit LSTM model requires significantly less energy (0.009 mJ, over 40x lower) but yields 14.89% worse accuracy (MSE 0.0432) and much longer training time. This trade-off highlights the need to align model selection with deployment priorities, favoring LSTM for ultra-low energy consumption or Transformer for higher predictive accuracy. In general, our work enables local, energy-efficient forecasting, contributing to more resilient combined sewer systems. All code can be found in the GitHub Repository (https://github.com/tianheng-ling/EdgeOverflowForecast).

Problem

• Forecasting sewer overflow basin levels for early intervention
• Deploying AI models on edge devices during communication outages
• Balancing energy efficiency and accuracy in embedded forecasting

Innovation

• Lightweight Transformer and LSTM models integration
• Integer-only quantization for efficient execution
• Automated hardware-aware deployment pipeline optimization