Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

HyperSense: Hyperdimensional Intelligent Sensing for Energy-Efficient Sparse Data Processing

Sanggeon Yun, Hanning Chen, Ryozo Masukawa, Hamza Errahmouni Barkam, Andrew Ding, Wenjun Huang, Arghavan Rezvani, Shaahin Angizi, Mohsen Imani

TL;DR

The proposed HyperSense model combines high‐performance software for object detection with real‐time hardware prediction, introducing the novel concept of intelligent sensor control, and is positioned as a promising solution for intelligent sensing and real‐time data processing across diverse applications.

Abstract

Introducing HyperSense, our co-designed hardware and software system efficiently controls Analog-to-Digital Converter (ADC) modules' data generation rate based on object presence predictions in sensor data. Addressing challenges posed by escalating sensor quantities and data rates, HyperSense reduces redundant digital data using energy-efficient low-precision ADC, diminishing machine learning system costs. Leveraging neurally-inspired HyperDimensional Computing (HDC), HyperSense analyzes real-time raw low-precision sensor data, offering advantages in handling noise, memory-centricity, and real-time learning. Our proposed HyperSense model combines high-performance software for object detection with real-time hardware prediction, introducing the novel concept of Intelligent Sensor Control. Comprehensive software and hardware evaluations demonstrate our solution's superior performance, evidenced by the highest Area Under the Curve (AUC) and sharpest Receiver Operating Characteristic (ROC) curve among lightweight models. Hardware-wise, our FPGA-based domain-specific accelerator tailored for HyperSense achieves a 5.6x speedup compared to YOLOv4 on NVIDIA Jetson Orin while showing up to 92.1% energy saving compared to the conventional system. These results underscore HyperSense's effectiveness and efficiency, positioning it as a promising solution for intelligent sensing and real-time data processing across diverse applications.

HyperSense: Hyperdimensional Intelligent Sensing for Energy-Efficient Sparse Data Processing

TL;DR

The proposed HyperSense model combines high‐performance software for object detection with real‐time hardware prediction, introducing the novel concept of intelligent sensor control, and is positioned as a promising solution for intelligent sensing and real‐time data processing across diverse applications.

Abstract

Introducing HyperSense, our co-designed hardware and software system efficiently controls Analog-to-Digital Converter (ADC) modules' data generation rate based on object presence predictions in sensor data. Addressing challenges posed by escalating sensor quantities and data rates, HyperSense reduces redundant digital data using energy-efficient low-precision ADC, diminishing machine learning system costs. Leveraging neurally-inspired HyperDimensional Computing (HDC), HyperSense analyzes real-time raw low-precision sensor data, offering advantages in handling noise, memory-centricity, and real-time learning. Our proposed HyperSense model combines high-performance software for object detection with real-time hardware prediction, introducing the novel concept of Intelligent Sensor Control. Comprehensive software and hardware evaluations demonstrate our solution's superior performance, evidenced by the highest Area Under the Curve (AUC) and sharpest Receiver Operating Characteristic (ROC) curve among lightweight models. Hardware-wise, our FPGA-based domain-specific accelerator tailored for HyperSense achieves a 5.6x speedup compared to YOLOv4 on NVIDIA Jetson Orin while showing up to 92.1% energy saving compared to the conventional system. These results underscore HyperSense's effectiveness and efficiency, positioning it as a promising solution for intelligent sensing and real-time data processing across diverse applications.
Paper Structure (24 sections, 12 equations, 17 figures, 3 tables)

This paper contains 24 sections, 12 equations, 17 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. HyperSense Model Design
  4. HDC Basics
  5. HDC Intelligent Sensor Control Framework
  6. HDC Object Detection
  7. Accelerator Architecture Design
  8. Computation Bottleneck
  9. HDC Encoding Optimization
  10. Accelerator Architecture Design
  11. Computation reuse example
  12. Experiments
  13. Experimental Setup
  14. Evaluation of HyperSense model
  15. Fragment model performance
  16. ...and 9 more sections

Figures (17)

  • Figure 1: Existing sensing and information processing pipeline.
  • Figure 2: Characteristics HDC possesses while DNN lacks. These characteristics of HDC make our intelligent sensing more powerful.
  • Figure 3: Overview of our Intelligent Sensing pipeline.
  • Figure 4: Illustration of how our proposed framework disables ADC to prevent the excessive generation of digital frames. Generating digital frames with no useful information cause an unnecessary increased cost in the system without HyperSense.
  • Figure 5: Overview of our object detection framework for Intelligent Sensing. The object detection framework consists of two models: (a) Fragment model and (b) HyperSense model. The trained Fragment model is applied to the HyperSense model.
  • ...and 12 more figures