Using an Extended Anomaly Model to Provide RSVP-Based QOS in Wireless-Lans

The anomaly of WLAN causes a lot of challenges concerning quality of service (QoS) on the wireless link and thus constricts the rollout of new applications and services. The anomaly means that the effective throughput of a WLAN client always depends on the other clients in its cell. This paper describes how to predict the throughput in a WLAN cell based on the number of clients in this cell and their specific automatic rate selection (ARS) rates. Therefore an extended anomaly model is used to compute the throughput at the time. Afterwards it is shown in which way the computed maximum throughput can be used in conjunction with the classical RSVP mechanisms to ensure QoS. The approach is evaluated by a simulation and by a Linux-based implementation in the Wireless Lab as well. Our approach allows providing voice over WLAN (VoWLAN) services even if the mobile clients do not support standardized QoS mechanisms like IEEE 802.11e