Research Of Algorithm Optimization On Wireless Access In WLAN

In recent years, wireless LAN (Wireless Local Area Networks, referred to as WLAN) has been most widely used to provide wireless association technology for wireless equipments. Network service providers are also increasing the number of access points (referred to as AP) gradually, to provide better wireless access and user experience. Among those wireless technology, good AP access technology is the key to ensure reasonable and efficient use of networks.Firstly, for improving handoff latency for AP access, this paper has studied handoff scheme in wireless mesh networks. When users need to trigger a handoff process, they often need to disconnect the existing connected AP and scan other APs in their range. Therefore, this paper proposes a handoff algorithm based on caching list. The core of the algorithm is that users cache a list of available APs around, when users need to switch the connected AP, they try to connect with APs in the caching list firstly. If there is not any available AP in the list, then active scan will be performed. This will greatly reduce handoff latency.After the research of fast handoff, this paper continues studying the wireless access with practice, taking the environment of campus into account, collecting the students’network usage in classrooms in zone B, proposing sociality-aware AP selection algorithm based on spectral clustering. Meanwhile, this paper analyzes the data from different points. That is, by collecting statistics of the network usage, relationships of users are computed by spectral clustering. Then through the proposed AP selection algorithm, users are allocated to proper APs.Finally, wireless access algorithms proposed in this paper can achieve the goal of optimizing network performance. Fast handoff based on caching list can significantly reduce handoff latency and improve the quality of service. This paper also simulates the sociality-aware AP selection algorithm based on spectral clustering. The results show that the proposed algorithm can improve the load balancing by at least 22% when compared with the least load first algorithm and 58% compared with random selection algorithm.