Application specific processors for the autoregressive signal analysis

Abstract

Two structures of the processors for the autoregressive analysis are considered. The first of them implements the Durbin algorithm using the rational fraction calculations. Such calculations provide higher precision than integer calculations do, and are simpler than the floating point calculations. The second of them implements the adaptive lattice filter. These processors are configured in FPGA, and give the possibility of the signal analysis with the sampling frequency up to 300 MHz. They provide new opportunities for the real time signal analysis and adaptive filtering in radio receivers, ultrasonic devices, wireless communications, etc. © 2010 Springer-Verlag Berlin Heidelberg.