Research Of TD-SCDMA Frequency Assignment Mechanism Based On Hyper-heuristic Genetic Algorithm

With the development of the mobile communications industry,especially since the development of mobile internet,mobile communication traffic has increased dramatically,communication services like mobile social network,mobile payments have become essential basic services for people.However,the frequency resources which can be used to carry mobile signal are very limited.Contradictory between limited frequency resources and surge of capacity serious impediment to the further development of mobile communications networks.To take full advantage of frequency resources,improve the quality of mobile communication networks,the frequency assignment problem must be solved.The traditional methods for deterministic method to solve the problem of frequency assignment problem such as pruning method is time consuming,within the effective period of time is impossible to obtain the solution of the problem;meta-heuristic such as simulated annealing,face premature convergence and get greater frequency interference,lower-quality solution.Against the traditional algorithm's defects in quality and time,this thesis proposes a hyper-heuristic genetic algorithm to solve the frequency assignment problem in TD-SCDMA.First of all,in order to improve the quality of frequency assignment,this thesis designs a hyper heuristic genetic algorithm to solve the frequency assignment problem.In this algorithm,six kinds of low-level heuristics were designed,as well as high-level genetic algorithm.On this basis,using network's point to view frequency assignment problem,this thesis proposes a strategy to assign the important cells first and then assign the unimportant cells to solve frequency assignment problem.This strategy can save much time and improve the efficiency of the process.Secondly,when using hyper heuristic genetic algorithm to solve frequency assignment problem,some cells easy to fall into local optimum,low-level heuristic strategy selecting these cells to reassign frequency would lead t o invalid search.To avoid this situation,this thesis designs a tabu search to assist low-level heuristic strategy to skip these cells to improve the efficiency of the algorithm.Finally,when using hyper heuristic genetic algorithm to solve the frequency assignment problem of large-scale or super large-scale network would consume much time.In order to save time and improve the efficiency of frequency assignment problem,this thesis proposes a regional division based frequency assignment method.In this method,service area is divided into two parts by a communal area,and then hyper heuristic genetic algorithm was used to assign frequency of the two part area in parallel method.The results of the experiments on two actual TD-SCDMA networks shows that hyper heuristic genetic algorithm can get better than existing methods in the quality of frequency assignment,and regional division based frequency assignment method can ensure the premise of quality,greatly accelerate the time consuming of frequency assignment.This thesis studies the frequency assignment problem in the aspects of quality and time.Against the traditional algorithm's defects in these two aspects,this thesis respectively designs a hyper heuristic genetic algorithm and a regional division b ased hyper heuristic genetic algorithm to solve frequency assignment problem in TD-SCDMA.The experimental results show that the algorithm giving consideration to frequency allocation quality and execution time spending,obtain the better frequency assignment scheme.The superiority in quality and time has important significance to solve the frequency assignment problem in real life.