Overhead in available bandwidth estimation tools: Evaluation and analysis

Abstract

Current Available Bandwidth Estimation Tools (ABET) insert into the network probing packets to perform a single estimation. The utilization of these packets makes ABET intrusive and prone to errors since they consume part of the available bandwidth they are measuring. This paper presents a comparative of Overhead Estimation Tools (OET) analysis of representative ABET: Abing, Diettopp, Pathload, PathChirp, Traceband, IGI, PTR, Assolo, and Wbest. By using Internet traffic, the study shows that the insertion of probing packets is a factor that affects two metrics associated to the estimation. First, it is shown that the accuracy is affected proportionally to the amount of probing traffic. Secondly, the Estimation Time (ET) is increased in high congested end-to-end links when auto-induced congestion tools are used.