Design of SOC Front End for Sensor Interface "Implementation of Compressive Sensing on ECG signals Processing SoC's"

  • Hamza Y. Al-Maharmeh

Student thesis: Master's Thesis


The goal of this project is to explore, design and develop an appropriate hardware implementation of Compressive Sensing (CS) for IoT devices. This research targets ultra-low power IoT applications such as implantable or wearable medical devices. CS can be used to acquire the data at much lower rate compared to conventional Nyquist rate. In Nyquist/Shannon sampling theorem, in order to recover the signal without loss, the sampling frequency has to be at least twice the highest frequency of the signal. In contrast, CS can sample the data at much lower rate while maintain excellent quality of the recovered signal. However, using conventional CS we are still limited with the maximum compression ratio that we can achieve without affecting the quality of the reconstructed signal. To address the problems with the existing ECG acquisition techniques, we propose the following enhancements: an adaptive CS (ACS) algorithm that combines Nyquist sampling theorem and Compressive Sensing, the combined Nyquist rate and CS algorithm can achieve higher compression ratios when compared to conventional CS techniques and results in smaller memory size requirements, due to the smaller number of samples the proposed algorithm can achieve lower power consumption than existing state of the art techniques, and finally the proposed algorithm is suitable for inclusion in IoT systems where ultra-low power is a necessity for many applications. The simulation results show excellent performance of ACS algorithm at higher compression ratios (CR=4.16), which is far superior compared to CR=2.5 using traditional CS. Finally, an ASIC design of the proposed algorithm is presented. Seven reconstruction algorithms have been studied to reconstruct the CS signal: Orthogonal Matching Pursuit (OMP), Simultaneous Orthogonal Matching Pursuit (SOMP), Regularized Orthogonal Matching Pursuit (ROMP), Compressive Sampling Matched Pursuit (CoSaMP), Spectral projected gradient (SPGL1), Smoothed L0 norm (SL0), and cspocs_K. The simulation results of these CS reconstruction techniques are presented in addition to many metrics that were used to evaluate performance and quality of reconstruction such as Signal to Noise Ratio (SNR), Peak Signal to Noise Ratio (PSNR), Mean Squared Error (MSE), Mean Absolute Error (MAE), Percentage Root Mean Square Difference (PRD), and Cross Correlation/Correlation Coefficient (CC). This report is organized as follows: an introduction to Compressive Sensing for IoT applications. Next a discussion of the recent research about compressive sensing and theoretical concepts behind it. Then different compressive sensing methods are presented. After that, the Adaptive Compressive Sensing (ACS) algorithm is discussed along with simulation results. Also, a detailed the RTL implementation of Adaptive CS algorithm is presented. Finally, the work is concluded.
Date of AwardJul 2017
Original languageAmerican English
SupervisorHani Saleh (Supervisor)


  • Compressive Sensing
  • ECG signal
  • Adaptive CS
  • CS hardware
  • SL0 Verilog
  • compressive sensing RTL.

Cite this