A. Ebrahimzadeh and A. Ranjbar (Iran)
Signal type recognition, genetic algorithm, higher order moments, higher order cumulants, self-adaptive step-size learning algorithm.
An expert system is proposed for the problem of recognition of digital signals. In this system, a neural network with self-adaptive step-size learning algorithm is proposed for determination of class of signal based on the extraction the prominent characteristics. Higher order statistics are used as the impressive characteristics of the signals. In order to select the best features, we have used a genetic algorithm. Experimental results show that the proposed system has a percentage of correct recognition higher than 93% for SNR>3dB.
Important Links:
Go Back
