Convergence-Time Analysis for the HTE Link Quality Estimator

Abstract

Evaluating wireless links is a common task for many control mechanisms. However, the inherent variability of those estimates negatively impacts network performance. To reduce this variability, the Hypothesis Test Estimator (HTE) was recently developed as an alternative to the commonly employed moving averages. Performance analyses carried out in recent works found that HTE returns more stable estimates at the cost of a typically larger average estimate error. This work uses numerical simulations to complement the previous analyses, but now under the perspective of convergence time –i.e., how long it takes for actual changes in the link quality to be reflected in the estimates. Our results indicate that HTE has, in general, a better convergence time than the moving averages. They also show that further improving HTE's convergence time is not trivial, as simple variations of the method that aim to improve convergence do not result in significant gains.