指数退避算法

在看NDN的默认转发策略BestRoute Strategy中提到了指数退避算法,回忆了一下,即为:

在一个共享信道的情况下,当网络上的节点在发生冲突时,每个节点节点等待一定的时间后重新发送。在二进制指数退避算法中,等待时间随着以二为底的指数增长。如果重试失败,那么下次的等待时间将会是上次的等待时间二倍。如果重试次数大于最大重试次数,那么包将从包队列中去除。

BestRoute Strategy:把Interest发给具有最小cost的下一跳,没收到要重传时选择次小cost的下一跳

This strategy forwards a new Interest to the lowest-cost nexthop (except downstream). After that, if consumer retransmits the Interest (and is not suppressed according to exponential backoff algorithm), the strategy forwards the Interest again to the lowest-cost nexthop (except downstream) that is not previously used. If all nexthops have been used, the strategy starts over with the first nexthop.

This strategy returns Nack to all downstreams with reason NoRoute if there is no usable nexthop, which may be caused by: (a) the FIB entry contains no nexthop; (b) the FIB nexthop happens to be the sole downstream; © the FIB nexthops violate scope.

This strategy returns Nack to all downstreams if all upstreams have returned Nacks. The reason of the sent Nack equals the least severe reason among received Nacks.