Increasing the Energy-Efficiency of Wearables Using Low-Precision Posit Arithmetic with PHEE

TL;DR

This work addresses the energyConstraints of wearable biomedical devices by evaluating low-precision posit arithmetic as an alternative to traditional floating-point formats. It demonstrates that 16-bit posits can match 32-bit IEEE 754 accuracy in cough detection, while 8–10 bit posits are sufficient for robust R-peak ECG detection, offering substantial energy savings. The authors introduce PHEE, a modular architecture that integrates the Coprosit posit coprocessor with the X-HEEP/RISC-V platform via CV-X-IF, and provide ASIC-level area and post-synthesis energy results showing up to 38% area reduction and up to 54% energy savings at the functional unit level. These findings highlight the practical potential of low-precision posit arithmetic to extend wearable battery life without compromising diagnostic performance.

Abstract

Wearable biomedical devices are increasingly being used for continuous patient health monitoring, enabling real-time insights and extended data collection without the need for prolonged hospital stays. These devices must be energy efficient to minimize battery size, improve comfort, and reduce recharging intervals. This paper investigates the use of specialized low-precision arithmetic formats to enhance the energy efficiency of biomedical wearables. Specifically, we explore posit arithmetic, a floating-point-like representation, in two key applications: cough detection for chronic cough monitoring and R peak detection in ECG analysis. Simulations reveal that 16-bit posits can replace 32-bit IEEE 754 floating point numbers with minimal accuracy loss in cough detection. For R peak detection, posit arithmetic achieves satisfactory accuracy with as few as 10 or 8 bits, compared to the 16-bit requirement for floating-point formats. To further this exploration, we introduce PHEE, a modular and extensible architecture that integrates the Coprosit posit coprocessor within a RISC-V-based system. Using the X-HEEP framework, PHEE seamlessly incorporates posit arithmetic, demonstrating reduced hardware area and power consumption compared to a floating-point counterpart system. Post-synthesis results targeting 16nm TSMC technology show that the posit hardware targeting these biomedical applications can be 38% smaller and consume up to 54% less energy at the functional unit level, with no performance compromise. These findings establish the potential of low-precision posit arithmetic to significantly improve the energy efficiency of wearable biomedical devices.

Figures (8)

• Figure 1: Posit format showing the sign, regime, exponent, and fraction fields.
• Figure 2: Accuracy and dynamic range of 16-bit arithmetic formats. The dynamic range of bfloat16 exceeds the limits of the plot.
• Figure 3: ROC curve for each arithmetic together with its AUC and the FPR for a TPR of 0.95.
• Figure 4: F1 score for BayeSlope executed with different arithmetic formats.
• Figure 5: Accuracy and dynamic range of FP16, posit12, and posit10. The dynamic range of the posit formats is larger than that of FP16, although they have fewer significant bits.