Path capacity estimation is used for e.g admission control decisions and rate adaptation. This has resulted in a variety of proposed techniques, mostly for the wired domain. In this paper, we evaluate the accuracy of one packet-pair estimator in a wireless test-bed. We study the performance of the estimator when it is configured to be low-intrusive and also gain experience with its use on different radio transmission technologies and in the presence of traffic-shapers and IP packet fragmentation.
The results show that the estimator can give valuable information to an admission control element. However, in a wireless network operating on a contention based shared channel, the packet-pair technique is vulnerable to the scheduling order of the probe packets on multi-hop paths.We also conclude as expected that the estimator can not be used uncritically on all types of transmission technologies.