| Summary: | 碩士 === 國立雲林科技大學 === 電機工程系碩士班 === 94 === IEEE 802.11 wireless Local Area Networks (LANs) have been widely deployed as a mainstream connectivity solution for a broad range of users nowadays. Wireless LANs provide users in public venues such as campuses, enterprises, or Internet Cafés, with a convenient means to access the Internet. An important issue in IEEE 802.11 networks is that user concentration at few access points (APs) imposes uneven load on APs, causing degradation of overall system efficiency. This is because typically each station selects to associate with an AP providing the best signal strength, without taking bandwidth share with other contending stations into account.
As a remedy, this thesis presents an approach to balancing traffic load of APs in a two-leveled fashion. For an overloaded AP, its immediately neighboring APs are viewed to comprise the inner layer (centered around the overloaded AP); next closest APs in the neighborhood form the outer layer. When a load balancing procedure is activated, each of inner-layer APs notifies its corresponding outer-layer APs of this activity. In response, involved APs select candidate stations from their maintained topology table in light of throughput and received signal strength measures. Assuming that selected candidate stations would actually be switched across APs, a resulting load balance index β is calculated as well. Then outer-layer APs report newly calculated results back to the overloaded AP. With knowledge of new β’s, the overloaded AP resolves the best-fit station for disassociation. The resolved station will associate with a new AP within radio range shortly. The association may cause the new AP to hand some local station off to another AP accordingly for load balancing purpose. Therefore our design leads APs to be better balanced in local scale to gradually global extent. Simulation results show that the proposed approach contributes to an improvement of 3.97% to 13.3% in terms of load balance index. Further, our approach accounts for an increase of throughput of 12.41% to 34.2%, a marked improvement. Evaluation results indicate that our proposal facilitates operations of the whole network system.
|
|---|
