Research And Application Of GSM Frequency Planning And Optimizing Based On Genetic Algorithms

Frequency planning and optimizing is one of the most important works in the network optimization. The frequency assignment problem, which is based on the channel assignment problem (CAP), belongs to one type of classical combinatorial optimization problem. It refers to assigning a limited number of available channels to many mobile voice or data users while minimizing electromagnetic interferences. This paper presents genetic algorithm-based solutions to frequency allocation in GSM networks.Firstly, we describe a CAP model, which uses a compatibility matrix to describe interferences among cells in the mobile network. We define a two-dimensional matrix F in the form of to encode the solution and regard the total constraints-violated times in F as the fitness function. Then we present a genetic algorithm for CAP. This algorithm uses modified roulette wheel selection strategy, two-point crossover and selected mutation, etc. genetic operators, and gets the satisfactory frequency assignment result that has lower fitness values. The effectiveness of this solution has been verified by the 21-cell benchmark.However, with respect to GSM networks, it is difficult to obtain the channel compatibility matrix from the interferential relationships among the cells, and there are also some engineering-oriented constraints that can not be expressed in terms of the compatibility matrix. So we put forward the algorithm GFA (Genetic Frequency Assignment). GFA takes the co-channel interference matrix and the adjacent channel interference matrix instead of the compatibility matrix to describe the interferential relationships among the cells. It defines the total cost of interferences in these two matrixes as the fitness function, and improves the initialization scheme for the algorithm. GFA also adopts the mini-adjusting strategy to satisfy the engineering-oriented constraints.Finally, according to the algorithm GFA, we develop the frequency optimizing software, by means of the programming language C#, the geographical information system MapXtreme 2004 and SQL Server 2005. This frequency optimizing software has been applied to optimize the GSM networks of China Mobile in Yunfu and Huizhou of Guangdong Province, and has got good optimizing effects.