Floating point hardware and software realization for adaptive FIR fetal ECG estimation

This paper presents realizations of IEEE-754 single precision floating point non-invasive fetal ECG estimation based on QR Decomposition Recursive Least Square algorithm (QRD-RLS). Experiments of the system, which is implemented on Xilinx Zynq SoC platform, are carried out with electrocardiogram (ECG) data and the results with analysis are presented. The embedded system design aims for hardware optimization by saving arithmetic resources, streaming pipeline performance and software-aid computation and integration. The challenge of exploiting the system full potential with pseudo-parallel computation on multiple fetal ECG data packets is also examined.