An Experimental Evaluation of ACK-based Passive Bandwidth Estimation Methods in Ad Hoc Networks

The primary challenge in bandwidth estimation within ad hoc networks lies in achieving a balance between accuracy and overhead. Passive and model-based methods for bandwidth estimation have not witnessed significant advancements in recent years, primarily due to two reasons. Firstly, these methods necessitate sampling diverse information at the physical layer and link layer, imposing stringent requirements on sensor quantity and accuracy, thereby making it challenging to control the overhead. Secondly, accurately estimating channel state proves to be exceedingly difficult, with both types of methods exhibiting their limitations. In this paper, we introduce three passive ACK-based engineering-oriented bandwidth estimation methods: Westwood, BED, and BED-R. These methods refrain from assuming any underlying protocol stack conditions while estimating available bandwidth based on ACK timestamps and packet sizes. Finally, we evaluate the performance of the three methods on the openwifi hardware platform. The experimental results demonstrate that BED-R exhibits superior performance, followed by BED in second place, while Westwood performs comparatively worse.