Performance Analysis of Available Bandwidth Estimation Algorithm Based on EWMA and Kalman Filter

This paper proposes an algorithm for estimating available bandwidth of network nodes. The algorithm, based on statistical information of the network nodes, calculates link utilization and estimates the available bandwidth through the exponentially weighed moving average (EWMA) filter, and can thereby estimate precisely the available bandwidth of the link in the next time interval. Compared with conventional available bandwidth estimating algorithms, our algorithm is based on passive probing, rather than active probing, of statistical information of the network nodes. The results from simulation show that our algorithm estimates the available bandwidth more precisely and features simpler parameter control than do estimating algorithms using Kalman filters. The estimating algorithm can be used in scenarios such as measuring and designing network routing protocols and optimizing queue scheduling algorithms, thus has high real world value.