Application Of Genetic Algorithms In Frequency Assignment Issue

With the rapid development of communication technology,how to do effective frequency assignment to improve the spectrum efficiency has been a main problem in the spectrum management filed.In military application,with the widely use of electronic devices,the electromagnetic environment has become quite complex on the modern battlefield.If the frequency assignment is not reasonable,a variety of devices may have mutual interference,even they would paralyzed in serious conditions. Therefore,a reasonable spectrum management is an important guarantee to win the modern warfare.In addition,in civilian application,the contradiction between the growth of communication services and the crowded frequency resources is also increasingly sharp.Therefore,to strengthen Spectrum management so as to use frequency resources effectively has become more and more important.In order to carry out a reasonable frequency assignment,this paper uses genetic algorithms(GA) to solve this problem.First,I analyze the basic principles of GA,and introduce some basic concepts and implementing techniques of the algorithm.And then,I give the mathematical model of the frequency assignment problem,and deduce the calculation method of the objective function.The goal is to find an assignment method whose interference meets the requirements of the system.Then,I analyze the basic principles of the four level forecast model which is used to judge is there any EMI between two radios.In this paper,I use Symbol encoding method in the GA application.I number the available frequency and distribute the frequency ID to radios,and then I get a string of individual coding.The parameters of the radios are saved in a database.Combined with the frequency ID distributed to the radios I can evaluate the interference value of the individual.And then,I choose roulette selection,single-point crossover and simple mutation as genetic manipulation.I choose calculation time as the convergence condition,and save the outstanding individuals in each iteration process.After simulation I find that GA solved the frequency assignment problem.The whole algorithms are programmed in VB language.And we programme the spectrum management software.Finally,in order to improve the algorithms performance,I optimize the parameters of GA.The original parameters such as population size,crossover probability and mutation probability values are based on the experience values.Consequently, optimization must be carried out,if you want the algorithms in its best effectiveness.At last,compared to the pre-optimization algorithms,the optimization algorithm's computing speed is improved nearly 60%and its stability is improved 4 fold.Since the genetic algorithm needs a large number of individuals,in order to further improve the performance of the algorithm,this article concludes a parallel computing method of GA in the end.