Adaptive filtering method and related device

A data processing device for processing an input signal which is a corrupted form of an original signal, comprising:

• a filtering circuit having adjustable filter coefficients for filtering the input signal and creating an output signal;
• an error calculator that is configured to provide an error signal from the output signal;
• a step gain calculator that is configured to provide a step gain of a form 2K, K being of the form K1+K2 with K1 being a negative integer and K2 being a positive integer and K being adjusted based on a magnitude of the error signal and on a step gain parameter;
• a memory circuit for storing values of K2 for different values of the magnitude of the error signal;
• a sign unit that is configured to provide a polarity of the error signal;
• an updating circuit that is configured to:
• provide, for at least a specific one of the filter coefficients, an updating amount from a product of:
• a first term that is based on the input signal,
• a second term that is based on a polarity of the error signal, and
• the step gain, and
• adjust the specific one of the filter coefficients using the updating amount;
• and
• output the output signal that better approximates the original signal for further processing.

A storage medium comprising a software module for storing a set of instructions executable under control of a processor and provided for performing a method of processing a digital input signal which is a corrupted form of an original signal, the method comprising:

• filtering the input signal to create an output signal using a filter having adjustable filter coefficients;
• deriving an error signal from the output signal;
• deriving, for at least a specific one of the filter coefficients, an updating amount from a product of:
• a first term that is based on the input signal,
• a second term that is based on a polarity of the error signal, and
• a step gain of form 2K, K being of the form K1+K2 with K1 being a negative integer and K2 being a positive integer and K being adjusted based on a magnitude of the error term and on a step gain parameter;
• storing values of K2 for different values of the magnitude of the error signal;
• adjusting the specific one of the filter coefficients using the updating amount;
• and
• outputting the output signal that better approximates the original signal for further processing.