Since the working nodes must be identified before new paths can be
routed, incoming calls are processed in batches. When a new call
originates at an input, it waits until the paths are established for
the batch that is currently being processed. When all of the calls in
that batch have been established, the working nodes are identified,
and then the paths for the new batch are established. Since
identifying the working nodes and routing the new paths both take
at most 
bit steps, the time to process each batch is
bit steps, and no call waits for more than 
bit
steps before being established, including the time waiting for the
previous batch to finish.