Study On Load Balancing Of Wireless Local Area Network

Studies have shown there are some issues in actual wireless LAN, such as load balancing and redundant data transmission and so on. They have a serious impact on network performance and will also result in unfair bandwidth allocation among users equality issues. In this article, we collected more than10,000users’historical data in a wireless LAN, through in-depth data analysis we found that user network behavior (such as co-leaving behavior of wireless networks) are often significantly affect the net-work load balancing; while we also observed that mobile users who have a similar type of application interest will have a stronger likelihood of network behavior. Based on the above observations, we propose a mobile user behaviour aware and application interest of mining based wireless access point load balancing algorithm, the algorithm collects historical network behavior of mobile users to learn social relations between users and users’application interests, in order to achieve load balancing between the wireless ac-cess point. Through experimental simulation based on real data, we confirmed that our algorithm is stable and efficient. Compared with the existing scheduling algorithms, our algorithm can gain load balancing performance for more than40%. Additionally, we propose a new redundant elimination cache algorithm which learn historical data to predict the contributing redundant data blocks, dynamically select the redundant data block resides in the cache and allocate different size of cache to different users. Simu-lation results show that compared to traditional FIFO caching algorithms, our method can achieve more than20%traffic saving.