Adaptive FIR filter for frequency and power estimation of sinusoids

Abstract

A new frequency and power estimation method which detects frequencies of unknown numbers of source sinusoidal signals is suggested. The system contains an adaptive finite impulse response filter (FIR) which exploits one of several random search algorithms, particle swarm optimisation (PSO). PSO will work on minimising the overall output power by finding the frequencies of the roots on the unit circle. The pseudo-spectrum is achieved by frequency elimination of the roots and the real power spectrum is derived from the pseudo-spectrum and the filter polynomial. The frequency, real power, and number of source signals are all estimated from the real power spectrum. After estimating the frequencies of the undesired interfering signals, the system suppresses these signals while maintaining the strength of the desired signals.