Design and evaluation of vivado HLS-Based compressive sensing for ECG signal analysis

Abstract

This paper proposes an efficient implementation of the compressed sensing technique. Specifically, we present an architecture for implementing the Matching pursuit algorithm using the Vivado High-Level Synthesis Tool. The proposed architecture consists of compressing and recovering the electrocardiogram signal on a hardware. For the healthcare application, there is an obvious need to get the algorithm implementation up and running in real-Time. The proposed implementation is applied on two different blocks sizes of the input data. Our results suggest that this difference in the size of input data produces consistent performance at the speed of execution. Specifically, according to our simulation, the blocks size of 256 samples achieve the speed by 8X as compare of a block of 2000 samples. Therefore, a calculation time of 2000 samples was 598ms and the computing time of 256 samples was 8.61ms and the necessary time to compute 2000 samples using a block of input 256 data was 67.2ms, which brings back to use a small vector to expect the best performance in terms of speed and area consumption. The hardware implementation results were compared with software implementation and the error between these two implementations was negligible of an order of 10-15