Thus packets are aggregated
opportunistically on their way to the sink.
When the sink receives the packets, it can decide to reinforce one or more paths
to transmit data at a higher data rate by sending a new interest with a shorter interval
back to the source along those paths, as shown in Figure 2(d). The decision of which
path to reinforce depends on the applications. The sink can choose those paths with
shorter delay, or paths from which more packets are received. Nodes receiving the
new interest update their gradients such that they can forward packets to the sink at
higher data rates, as shown in Figure 2(e).
Each interest has a duration field specifying how long the interest will be effective.
The gradient also contains this field to decide when the nodes should stop
reporting the events. Therefore the sink has to send interest to the reinforced paths to
update the gradients to maintain the paths. Un-reinforced paths will stop forwarding
packets to the sink after the gradient expires, as shown in Figure 2(f). This reduces
unnecessary duplicated packet transmissions. To further reduce unnecessary transmissions,
the sink can send negative-reinforcement messages to stop some paths
from forwarding packets.
Directed Diffusion can achieve robust communication by forwarding packets on
multiple paths that are selected by the reinforcement mechanism and are preferred
by the sink. When sensors sense an event, the event is reported to the sink as long
as there is at least one feasible path from the source to the sink.
Pages:
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534