Opportunistic multicast schemes for time-sensitive stream traffic are studied for cellular networks where erasure coded packets are transmitted over discrete-time quasi-static forward-link fading channels. Assume that to successfully decode a transmitted stream fragment, it suffices that one receives k packets from the fragment. Two important issues are thus raised on when to stop transmitting a fragment and how to minimize the stopping time (ST) through transmission rate scheduling. Based on available channel state information and scheduled history, we tackle them particularly for small k required for short latency. If the distribution of IID channel states is available, the scheme is to compute and use the optimal constant transmission rate and minimum fixed ST subject to a reliability constraint. We show that the minimum ST grows with the logarithm of multicast group size. If channel state information is available, we propose to minimize random ST through selecting each optimal instantaneous transmission rate for a utility function. The utility function is specifically designed to exploit system transient states and dynamics. Results show that the scheme with an exponential weighted residual work achieves the least mean ST but the scheme maximizing instantaneous effective sum throughput has an edge of low complexity.
ASJC Scopus subject areas