A semi-automatic configuration control system for high-density WiFi network in large indoor environment

• Main Authors: Zhi-Qian Zheng, 鄭智謙
• Other Authors: 蔡志宏
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/65zyyw

碩士 === 國立臺灣大學 === 電信工程學研究所 === 107 === In order to meet the needs of various applications in various large-scale fields such as factories or public spaces, enough access points(APs) must be built to ease the congestion of the network. However, in the case of limited spectrum resources, the quality of such wireless networks will not be improved with the increasing number of access points. On the contrary, as the density of access points in a fixed area becomes higher, the signals from different access points will cause interferences among different radio links. Without professional background, a high-overlap, high-interference wireless network environment may be generated, and data transmission rate or delay may be seriously degraded due to collisions and interference. To solve this type of problems, exhaustive methods or genetic algorithms are often used, but the time and computation are quite large. In order to improve overall system efficiency, shorten time and approximate optimal solution, this study proposes a semi-auto configuration control system(SCCS), which employs periodic adjustment of power and channel bandwidth of various APs. There are two main indicators for evaluating the performance of algorithm: (1) the number of devices meeting their quality of service and (2) overall system interference condition. Utility function is the multiple of two indicators for assessing overall performance. The experimental simulation considers a variety of different fields, and the most effective scenario for SCCS is the high-density field, which SCCS improve performance about 52% after optimization. When compared genetic algorithms with SCCS, genetic algorithms can achieve good results in most of the optimization problems, but the consumed computation time and huge amount of computation requirement are still the most significant limits in practice. Providing the balance of the computational time and the pros and cons of the results is the main contribution of this study.