Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Characterizing Accuracy Trade-offs of EEG Applications on Embedded HMPs

Zain Taufique, Muhammad Awais Bin Altaf, Antonio Miele, Pasi Liljeberg, Anil Kanduri

TL;DR

The paper tackles the challenge of running EEG analysis on energy-constrained wearables by leveraging embedded heterogeneous multi-core platforms (HMPs) and error-resilient approximations. It studies three real-world EEG applications—Epileptic Seizure Detection, Sleep Stage Classification, and Stress Detection—on an Odroid XU3, using Welch-based PSD features and a multi-parameter design space that combines approximation levels, DVFS, and core mapping. Through a 3-D Pareto analysis of power, performance, and accuracy, it demonstrates that higher approximation yields better performance at similar power and that the LITTLE cluster often offers superior energy efficiency compared to a single big-core configuration. These findings provide practical guidelines for disciplined approximation tuning to maximize performance within energy budgets for embedded EEG on HMPs.

Abstract

Electroencephalography (EEG) recordings are analyzed using battery-powered wearable devices to monitor brain activities and neurological disorders. These applications require long and continuous processing to generate feasible results. However, wearable devices are constrained with limited energy and computation resources, owing to their small sizes for practical use cases. Embedded heterogeneous multi-core platforms (HMPs) can provide better performance within limited energy budgets for EEG applications. Error resilience of the EEG application pipeline can be exploited further to maximize the performance and energy gains with HMPs. However, disciplined tuning of approximation on embedded HMPs requires a thorough exploration of the accuracy-performance-power trade-off space. In this work, we characterize the error resilience of three EEG applications, including Epileptic Seizure Detection, Sleep Stage Classification, and Stress Detection on the real-world embedded HMP test-bed of the Odroid XU3 platform. We present a combinatorial evaluation of power-performance-accuracy trade-offs of EEG applications at different approximation, power, and performance levels to provide insights into the disciplined tuning of approximation in EEG applications on embedded platforms.

Characterizing Accuracy Trade-offs of EEG Applications on Embedded HMPs

TL;DR

The paper tackles the challenge of running EEG analysis on energy-constrained wearables by leveraging embedded heterogeneous multi-core platforms (HMPs) and error-resilient approximations. It studies three real-world EEG applications—Epileptic Seizure Detection, Sleep Stage Classification, and Stress Detection—on an Odroid XU3, using Welch-based PSD features and a multi-parameter design space that combines approximation levels, DVFS, and core mapping. Through a 3-D Pareto analysis of power, performance, and accuracy, it demonstrates that higher approximation yields better performance at similar power and that the LITTLE cluster often offers superior energy efficiency compared to a single big-core configuration. These findings provide practical guidelines for disciplined approximation tuning to maximize performance within energy budgets for embedded EEG on HMPs.

Abstract

Electroencephalography (EEG) recordings are analyzed using battery-powered wearable devices to monitor brain activities and neurological disorders. These applications require long and continuous processing to generate feasible results. However, wearable devices are constrained with limited energy and computation resources, owing to their small sizes for practical use cases. Embedded heterogeneous multi-core platforms (HMPs) can provide better performance within limited energy budgets for EEG applications. Error resilience of the EEG application pipeline can be exploited further to maximize the performance and energy gains with HMPs. However, disciplined tuning of approximation on embedded HMPs requires a thorough exploration of the accuracy-performance-power trade-off space. In this work, we characterize the error resilience of three EEG applications, including Epileptic Seizure Detection, Sleep Stage Classification, and Stress Detection on the real-world embedded HMP test-bed of the Odroid XU3 platform. We present a combinatorial evaluation of power-performance-accuracy trade-offs of EEG applications at different approximation, power, and performance levels to provide insights into the disciplined tuning of approximation in EEG applications on embedded platforms.
Paper Structure (10 sections, 2 equations, 10 figures, 1 table)

This paper contains 10 sections, 2 equations, 10 figures, 1 table.

Table of Contents

  1. Introduction
  2. Error resilience in EEG Applications
  3. Welch Method
  4. Approximation Methods
  5. Characterizing Accuracy Trade-offs
  6. Datasets of EEG Applications
  7. Hardware Setup
  8. Implementation Strategy
  9. Evaluation and Results
  10. Conclusion

Figures (10)

  • Figure 1: The Performance gain vs accuracy loss in Welch using window overlap method (a) Accurate, 50% overlap, (b) Level-1 approximation, 25% overlap, (c) Level-2 approximation, no overlap, (d) Performance gain vs Approximation
  • Figure 2: Performance characteristics at 600 Hz for (a) Epilepsy, (b) Sleep, (c) Stress.
  • Figure 3: Power characteristics at 600 Hz for (a) Epilepsy, (b) Sleep, (c) Stress.
  • Figure 4: Accuracy trade-off vs power-performance characteristics at 600 Hz for (a) Epilepsy, (b) Sleep, (c) Stress.
  • Figure 5: Performance characteristics at 1000 Hz for (a) Epilepsy, (b) Sleep, (c) Stress.
  • ...and 5 more figures