An Application of Fuzzy Hopfield Neural Network on Communication Scheduling Problems
博士 === 國立成功大學 === 工程科學系碩博士班 === 95 === The demand for wireless communication service is increasing rapidly. But the available electromagnetic frequency spectrum is rigorously limited. How to manage the frequency resources effectively to provide the maximum service capacity becomes an important issue...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | en_US |
| Published: |
2007
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/87713513228810918035 |
| Summary: | 博士 === 國立成功大學 === 工程科學系碩博士班 === 95 === The demand for wireless communication service is increasing rapidly. But the available electromagnetic frequency spectrum is rigorously limited. How to manage the frequency resources effectively to provide the maximum service capacity becomes an important issue. This thesis will use the Fuzzy Hopfield Neural Network to probe the communication scheduling instances.
In recent years, Fuzzy Hopfield Neural Network is frequently used to solve optimization problems. The problems are mapped to a FHNN system and an energy function corresponds to the best solutions of the system is chosen. Then the optimal design procedures of the system are conducted. Two different communication scheduling examples, including channel assignment problem and broadcast scheduling problem are considered.
In this thesis, channel assignment problem is considered firstly. The cells on a cell phone communication system are considered as clusters and the frequency channels as objects or samples. The channel assignment problem is then as the optimal design problem of clustering the samples into fixed number of clusters. The channel assignment is an NP-hard problem. It is not realistic for solving the problem with analytic method. The second problem to be considered is the broadcast scheduling problem for packet radio networks. The BSP problem is also an NP-hard problem. Both of the above NP-hard problems are solved by the applied fuzzy Hopfield neural network. Simulation results show that the FHNN can provide an alternative approach for solving these class communication scheduling problems effectively.
|
|---|