SUMMARYOne of the main features of the upcoming IEEE 802.11s standard is the path selection procedure, which is performed by the Hybrid Wireless Mesh Protocol by means of thetextitairtime link metric (ALM). Through a small experimental testbed, and using the open80211s code, which provides one of the most compliant draft implementations, we analysed the impact that the implementation choices of the developers have on the system performance. The analysis revealed that even small deviations from the specified routines (or different choices in the estimation of the parameters used in the online calculation of the ALM) can lead to noteworthy performance degradations. We thus realized some fixes to improve the performance of the system and verified the effectiveness of such amendments in a realistic simulation environment. We then compared the performance of the modified version to the original one. The outcome of the test highlighted the remarkable advantage achieved by the proposed fixes. Copyright © 2011 John Wiley & Sons, Ltd.