And also if you know some new and good references please let me know, thank you so much

@Negar, I'm quite a bit puzzled - why OGA might be considered superior by any other reason than just convenience? could you pls. expound a little giving some background?

@Jan. first of all, thank you for your response

I know an OLGA is a GA, complete with a population of size p, fitness function, recombination operators, and mutation operators. However, unlike a traditional GA, the fitness of an individual changes over time, as it is exposed to more examples. now I want to know, how is better to use GA for signal processing(for example modulation classification), because of it's huge complexity in the components of signal, is it good and beneficial to obey fitness changes or apply traditional GA?

@Negar, several years back - perhaps at the time when OGAs were I-st introduces, but I'm not quite sure - some ideas to use them in a massively parallel mode, with a higher-lever voting ANN controller have been advanced, but I didn't quite follow them then. Do you know/considered whether something like this were ever fruitful?

@Jan, I've read some articles and studies about classification using OGA (university of New Jersey) with some merits, in those studies , authors claim that OGA better mimic the behavior of natural selection, as real organisms live in environments that are not identical to that of their ancestors, but for a special case like modulation classification, I'm not sure and need to study and search more

@Negar, yes, I'd agree with the natural evolutionists... IFF such optimization would be allowed to run in [quasi]infinite time - then the solutions achieved are perhaps bound to be nearly optimal !! And this is exactly where a massively parallel mode of operation, with a controlling supervisor pruning in accordance to some preset global cost function, might be advantageous. However, I'm not quite sure whether if a real-time operation is one of boundary conditions, there will be a sufficient processing power available to substitute massive parallelism for a [quasi]infinite evolution runs?

So, does the above describe in any way what you might be seeking for your modulation classifier?

@Jan , thanks a lot for your kindly attention and your guidance, yes certainly your answer help me so much, good luck