Abstract

The performance experienced by wireless clients in IEEE 802.11 wireless networks heavily depends on the clients' ability to identify the Access Point (AP) that will offer the best service. The current AP affiliation mechanism implemented in most wireless clients is based on signal strength measurements received by the client from all the APs in its neighborhood. The client then affiliates with the AP from which it receives the strongest signal. It is well-known that such an algorithm can lead to sub-optimal performance, due to its ignorance of the load at different APs.

In this work, we consider the problem of AP selection. We identify potential bandwidth as the metric based on which hosts should make affiliation decisions, and define it as the (MAC-layer) bandwidth that the client is likely to receive after affiliating with a particular AP. We further limit ourselves to the use of passive measurements that do not require an end-host to affiliate with the AP, thus allowing the end-host to simultaneously evaluate the potential bandwidth to multiple APs in range. This can also facilitate more informed roaming decisions. We propose a methodology for the estimation of potential upstream and downstream bandwidth between a client and an AP based on measurements of delay incurred by 802.11 Beacon frames from the AP. Preliminary experiments conducted in a controlled environment demonstrate that the proposed methodology looks promising, yielding fairly accurate results under varying conditions.