Experimental Analysis of the Channel Busy Time in Vehicular Ad-Hoc Networks

To maintain the benefits of fast growing Vehicular Ad-Hoc Networks (VANETs), Congestion Control (CC) algorithms are used to reduce channel congestion even under harsh communication conditions in order to achieve a reliable service. Reactive channel load (CL) based CC mechanisms, like Decentralized Congestion Control (DCC), therefore hardly rely on accurate CL values to work at best effort. While continuous probing of the channel obviously offers the best accuracy, it is not always possible or necessary, as illustrated in this paper. Instead estimation methods are used which may vary in accuracy through their parameter settings. The aim of this paper is to investigate the impact of synchronization and the measuring interval length for time (in-)variant CL as well as number and distribution of probes for the channel busy time. Various measurements, which are retrieved from an experimental set-up, are used for a theoretical evaluation. I illustrate that accuracy strongly varies and also suffers from the use of certain parameters, especially the measuring interval length and the synchronization effect which is of major interest for time (in-)variant CL occurring in VANETs. Furthermore the paper reveals that simple constant probing can, in terms of accuracy, outperform uniformly distributed probes used so far.