LEAST-SQUARES LATTICE ALGORITHMS FOR IIR NOTCH FILTERING

摘要

Adaptive notch filtering for sinusoids retrieval from background noise is an important subject in telecommunications world. This field is particularly interesting in applications such as radar, sonar and telecom systems that use variable carrier frequencies. This paper presents a new approach to the IIR notch filtering: recursive least-squares method applied to lattice filter structures represented through trigonometric functions. We have applied the Gauss-Newton method to gradient approaches and have developed two least-squares algorithms: a complete form and a simplified one. From the author's knowledge, the presented algorithms have never been published in these forms before. They are very good candidates for programming in CORDIC and implementation on dedicated FPGA processors. We present the developed methods and we compare their performances to the direct form RLS notch implementations.