Congestion relief in wireless LANs

The rapid proliferation of WiFi enabled devices is generating wireless traffic management challenges particularly at certain popular spaces ("hot-spots") within the network. We address two subsets of traffic engineering, namely load balancing and explicit channel planning amongst a group of access points (APs). The goal is to accommodate more users by optimizing capacity allocation, guarantee a minimum level of user experience and maximize resource utilization in the face of dynamic traffic patterns that change with location and time. We propose a procedure that minimizes an approximation to the weighted mean end-to-end delay experienced by wireless users that is attributable to the wireless portion of the network. We derive an expression for the cost function to be optimized and develop a two-step approach. In step 1, we nearly optimally assign stations to APs. In step 2, we optimally allocate frequencies to APs by formulating an optimal integer linear programming (ILP) problem. We show that step 1 in our algorithm improves average end-to-end delays in several two APs scenarios by 20% to 35%. It also supports over 50% more stations for the same mean end-to-end delay. The second step offers additional improvements of 40% in the overall mean end-to-end delay across multiple APs, as compared to other techniques that have been proposed or implemented.