An analysis of packet loss correlation in FEC-enhanced multicast trees

Reliable Rateless Wireless Broadcasting With Near-Zero Feedback

“…While in practice the packet loss probability differs from node to node due to many factors (i.e., link quality, distance to the source, antenna sensitivity), the works in [15], [17], [21], and [22], as well as some hybrid FEC/ARQ protocols such as [23], assume homogeneous packet loss probabilities in their analysis. The works in [16], [18]- [20] do study the more realistic scenario of heterogeneous packet loss, but they assume that receivers' packet loss probabilities are known to the source, a relatively strong assumption for practical multiple receiver environments. In our work, we allow the packet loss probabilities to be unknown and heterogeneous across receivers.…”

“…The goal of the ML estimator is to find the values of and that maximize the likelihood when given the observations . Namely (17) Equivalently, one can maximize the logarithm of the likelihood function, called log-likelihood, as following, (18) In order to obtain the estimation that maximizes the likelihood, one can set the partial derivatives of the log-likelihood function [(18)] with respect to and to zero. Therefore, using (15), one can obtain the ML estimation for and by solving the following system of equations: (19) Namely (20) Now back to the problem of estimating the extreme value index and the scale factor using ML estimator.…”

Network-embedded FEC for optimum throughput of multicast packet video

“…4 It is well known that losses over the Internet are bursty in nature (i.e., exhibit memory), and hence such losses do not necessarily follow a binomial distribution. Due to its simplicity though, many studies (including this paper) assumes a binomial distribution [9,10]. Evaluation of the performance of NEF under burst losses is part of our ongoing work.…”

Modeling multicast packet losses in wireless LANs

“…They calculated the loss covariance in each pair of traces and observed certain degree of spatial loss correlation among multicast sites. In [16], the impact of loss correlation is addressed. The authors studied group loss probabilities in shared loss multicast communications for the design of forward error correction algorithms.…”