FPGA implementation of respiration signal classification using a soft-core processor

This paper proposes an implementation of automatic classification of respiratory signals using a field programmable gate array (FPGA). It has been shown for this type of respiratory signal that second order autoregressive modeling (AR) combined with a modified zero-crossing algorithm results in close to 100% consistency between manual and automatic classification methods. This algorithm was improved by adding calibration procedures and adjusted to run on an FPGA. Altera\´s development tools and intellectual property (IP) mega-core functions were utilized to implement a "soft-core" processor capable of running compiled C algorithms inside the Stratii FPGA chip. In addition, the high density and flexibility of the FPGA allowed for coupling of the soft-core processor with other hardware modules to form a fast interface between off-chip devices. The external SRAM, flash memory, and an LCD were interfaced with the NIOS II soft-core processor through a system on a programmable chip (SOPC) design.