On-Device Soft Sensors: Real-Time Fluid Flow Estimation from Level Sensor Data
Tianheng Ling, Chao Qian, Gregor Schiele
TL;DR
Cloud-based soft sensors face latency and data-security challenges in real-time IoT contexts. The paper proposes an on-device soft-sensor framework by integrating a low-power MCU with an embedded FPGA (Elastic Node) to perform DL inference locally, with training on the cloud and deployment via TensorFlow Lite for MCU or ElasticAI.Creator for FPGA. Results show FPGA-based soft sensors achieving inference times in the range $1.04$ to $12.04\,\mu s$ and lower power than MCU, enabling real-time operation at 10 kHz, while MCU baselines are slower and more power-hungry; quantization-aware strategies significantly affect FPGA performance. Overall, the work demonstrates a practical, secure, low-latency edge-AI path for soft-sensor tasks and informs hardware-software co-design choices for embedded real-time perception.
Abstract
Soft sensors are crucial in bridging autonomous systems' physical and digital realms, enhancing sensor fusion and perception. Instead of deploying soft sensors on the Cloud, this study shift towards employing on-device soft sensors, promising heightened efficiency and bolstering data security. Our approach substantially improves energy efficiency by deploying Artificial Intelligence (AI) directly on devices within a wireless sensor network. Furthermore, the synergistic integration of the Microcontroller Unit and Field-Programmable Gate Array (FPGA) leverages the rapid AI inference capabilities of the latter. Empirical evidence from our real-world use case demonstrates that FPGA-based soft sensors achieve inference times ranging remarkably from 1.04 to 12.04 microseconds. These compelling results highlight the considerable potential of our innovative approach for executing real-time inference tasks efficiently, thereby presenting a feasible alternative that effectively addresses the latency challenges intrinsic to Cloud-based deployments.