FlowPrecision: Advancing FPGA-Based Real-Time Fluid Flow Estimation with Linear Quantization

TL;DR

FlowPrecision tackles real-time, on-device fluid-flow estimation when cloud processing is impractical. It proposes linear quantization with adaptive scalers in FPGA-based soft sensors to preserve neural network precision while remaining energy-efficient. The study demonstrates up to $10.10\%$ reduction in test loss and up to $9.39\%$ faster inference, validated on three data sets and deployed on Spartan-7. The results advocate FPGA-based, quantized inference as a viable, autonomous alternative to cloud-based processing for pervasive flow-monitoring systems.

Abstract

In industrial and environmental monitoring, achieving real-time and precise fluid flow measurement remains a critical challenge. This study applies linear quantization in FPGA-based soft sensors for fluid flow estimation, significantly enhancing Neural Network model precision by overcoming the limitations of traditional fixed-point quantization. Our approach achieves up to a 10.10% reduction in Mean Squared Error and a notable 9.39% improvement in inference speed through targeted hardware optimizations. Validated across multiple data sets, our findings demonstrate that the optimized FPGA-based quantized models can provide efficient, accurate real-time inference, offering a viable alternative to cloud-based processing in pervasive autonomous systems.

Figures (3)

• Figure 1: Quantization Description
• Figure 1: Data Set DS1: Measurements With Upward Trend
• Figure 2: Distribution of Test MSEs Across Various Hidden Neurons