When an IM will never
have a full batch and other competing IMs always have full
batches, the IM will never be served. To avoid the starvation
of that IM, if a full batch cannot be found for a certain
period, say 4n time slots, the VOMQ starts to pad the
VOMQ with dummy cells, one cell per time slot, until a full
batch is formed. Starting from then, the weight of the padded
VOMQ is calculated as the waiting time instead of the
queue length. Then at some time point, the waiting time
weight of the padded batch will be greater than other
VOMQs, so the VOMQ will finally be served.