Research On Load Balancing Strategy For Wireless Access Point Accessing Cloud Controller

WLAN (Wireless Local Area Networks) is expanding fast with the popularity of intelligent wireless terminal and the demand for wireless network. For the bottleneck problem of single AC(Wireless Access Controller) model, multi ACs are brought into the centralized WLAN network architecture. In the single AC system, AP(Wireless Access Point) is controlled by the single AC, so there is no need to care about the ownership of AP. But for the multi ACs system, especially the cloud controller system and the AC resource pool, the load balancing of AC cluster has become a problem to be solved. The existing AC cluster load balancing algorithm mainly focused on the number of APs, the number of users, the network traffic of ACs, as well as the CPU and memory usage. However, because of the network traffic burst characteristics and the imbalance of the user traffic on AP, only considering the current number of AP and the network traffic on ACs is not enough. Therefore, it is necessary to think about the load of ACs globally to reduce the imbalance network traffic of ACs.In this paper, the AC network traffic time series are reconstructed according to the Takens embedding theorem and the characteristic that the wireless network traffic is locally similar in days. Then it can predict for a long time with the reconstruction sequence. Based on the characteristics of APs accessing ACs, the study analyses several models, and selects an appropriate prediction model for AC network traffic. At last, this paper also discusses the coefficient of quadratic exponential smoothing model.Based on the characteristics of cloud controller system, a suitable load balancing strategy for centralized forwarding is proposed when APs accessing cloud controllers. This paper revises the least connection scheduling algorithm with the network traffic prediction. The revised least connection scheduling algorithm is simulated in MATLAB. At the same time, CloudSim is used to verify the effect of the minimum connection algorithm. The simulation results show that the revised least connection scheduling algorithm can improve the performance of reducing the load difference among ACs.