Study On Some Tropics For Mobile Communication System Of Guangdong Telecom

Some topics about Gunagdong Mobile Network, such as the development strategy, frequency planning are studied in this dissertation. Specially, the new theory of neural networks are applied for predicating subscribes and speech coding in the mobile network. The studied achievements in the dissertation are listed as following:1. Depending on the investigation and analyses of the network structure, the market, the service and some existing problems of the network, the development strategy and suggestions are presented.2. A new result on the function approximation of feedforward neural networks is proven, which indicates the equivalence of multi-demotion function approximation to the one dimensional function approximation, and thus simplifies the analysis of the function approximation of the feed forward neural networks. The mobile communication market of several cities of Guangdong is predicted used for reference in the planing of mobile communication of Guangdong province.3. Two new frequency reuse methods are presented. One is used in TACS network for reusing frequencies in a caster. The other is multiple reuse pattern for frequency planning in GSM network. The field experiences in Guangdong and Shenzhen networks show that the two methods can provide high capacity at low costs using existing macrocodes, without negatively affecting the network quality. They are two efficient frequency reuse patterns.4.Souce coding is a potential method to crease the capacity of mobile communication. A class of time-delay feedforward neural networks based on nonlinear perdition of speech signal is studied, and proposes the stability condition of the nonlinear predictor based on time delay neural networks. With the development of computer, integral, circuits, the method has large application potential.