Low Variance Loss Event Rate Estimation for Layered Multicast Protocol

In a high level of statistical multiplexing environment, the packet drop pattern of a data flow is determined by the aggregate data flow behaviour of the bottleneck link. However, an observation of loss event patterns of a layered multicast session shows oscillatory loss intervals are estimated at receivers. Under this condition, the current loss event rate estimation technique of layered multicast protocols estimates oscillatory loss event rates. Moreover, the protocols inability to synchronise packet sequence number across multicast layers exaggerates this problem. To address this problem, we propose two-step loss interval filtering technique that discards the unrepresentative loss interval samples. The proposed loss interval filtering technique reduces the variance of estimated loss event rates and results in stable protocol behaviour