the same IM but passed through different CMs. The OM
can easily control the order of these n cells since they arrive
at the same time. Then, each OM enqueues these n cells into
the OQs according to the cell’s destination output ports.
Some of the cells may be destined for the same output port,
and they are enqueued into the same OQ. This indicates the
need of shared-memory, which is not scalable. However, in
the next section, we will introduce the distributed sharedmemory
design to improve the scalability. At the same
time, during every time slot, each OQ sends one cell to the
corresponding ELC