| The goal of the mobile communication network planning and optimization is to guarantee the quality of services (QoS), network resources allocation has important impact on the QoS in networks. In accordance with the network operating and service conditions, adjusting and optimizing of various types of wired and wireless resources in mobile communication networks is an important part of network planning and optimization.The frequency in GSM networks and the scrambling code in TD-SCDMA networks are both important radio resources. Optimal allocation of frequencies and scrambling codes can effectively reduce the network interferences and improve the system capacity. Optimization of frequency and scrambling code allocation can be modeled as multi-objective and multi-constrained optimization problems, which need to balance and make trade off between more than one optimization objective and satisfy multiple constraints. Multi-objective evolutionary algorithms can effectively solve complex nonlinear multi-objective optimization problems. Genetic algorithms belong to the family of evolutionary algorithms and are suitable to deal with structural optimization problems. This thesis applies the multi-objective genetic algorithms (MOGA) to two complex nonlinear multi-objective optimization problems in mobile communication networks, that is:frequency allocation in GSM networks and scrambling code allocation in TD-SCDMA networks.Optimization of frequency allocation in GSM networks by means of MOGA is composed of three steps. At first, the objectives and constraints, which frequency allocation wants to achieve and is subject to respectively, assignment model is proposed. This model takes the minimization of the same frequency interference, adjacent frequency interference and overlapping degree as optimization sub-targets. Secondly, individual genes are encoded in the form of the assignment matrix, and the evolution of populations is processed by the crossover operator and mutation operator. The cross operator adopts the geometrical distance between individuals to determine the pairing of individuals, and the mutation operator dynamically adjusts mutation probability according to evolutionary generations. Then optimal individuals are selected into the Pareto optimal set, at the same time niches and the adaptive grid are used to maintain the uniform distribution and diversity of the populations. Finally, analytic hierarchy process is introduced to select a preference solution from the Pareto optimal set in order to provide the optimal frequency assignment plans in GSM networks.Optimal assignment of scrambling codes in TD-SCDMA networks is isomorphic to optimal allocation of frequencies in GSM networks. The multi-objective and multi-constrained model for scrambling code assignment is established, which takes the minimization of the same scrambling code interference, cross correlation scrambling code interference and overlapping degree as optimization sub-goals. Then in accordance with characteristics of the scrambling code optimization, some key technologies such as genetic coding, population initialization and population adjustment are illustrated. The other schemes that are similar to that in optimization of frequency allocation in GSM networks, for examples, fitness evaluation, selection, and mutation, are not described in details.Based on the key technologies above-mentioned, a prototype of optimal frequency-allocation is developed by means of the.NET platform, C#and SQL Server. Applying this system to the practical GSM network of a city in south China proves that it can quickly provide proper preference frequency assignment solutions.All above-mentioned works demonstrate that, MOGA based optimizing allocation of GSM frequencies and TD-SCDMA scrambling codes reflects a variety of optimization objectives reasonably, and can generate efficient assignment solutions rapidly, thus has good application prospects in practical network optimization. |
PDF Full Text Request