> Yes, I see. Then If the message is lost due to a collision, the node which > sends the message would not add > it's hearBeats in this send period, because hearBeats only represents the > number of sucessfully sent messages. true
> But another question is arised: whether there exsits this situation: the > message is lost due to some other reasons, > but the node dosen't know the transmition is failed, and the heartbeats still > adds itself. Could this happen? yes, this happens - for example if transmitters message collides with someone else's message (timesync messages are broadcasts, so ack wouldn't help). this shouldn't be a problem though, you want the heartbeats to advance in this case... brano > tanlin > > > > 发件人: Kusy, Brano > 发送时间: 2010-05-11 13:15:01 > 收件人: tanlin; tinyos-help@millennium.berkeley.edu > 抄送: > 主题: [?? Probable Spam] Re: Another question about FTSP > >> I'm not that certain about your paper "FTSP Protocol Verification using >> SPIN". >> In your paper, when modelling the Radio Channel Model, you said that the >> channel would drop a radio message with some probability. I'm wodering >> whether >> the node which transmits the synchronization message would know the loss of >> the message and re-transimit it. And will the node wait some time for >> another transmition when the radio collisions happen. > MAC protocol takes care of the channel sharing and should prevent radio > collisions for the most part. however, there is no retransmissions in FTSP. > sensornet deployments are typically quite dense - so each node should have a > multiple neighbors. the node will synchronize with other neighbors, if a > timesync packet is lost due to a collision. in the worst case, the node > waits until the next synchronization round. we haven't found this to be a > problem in practice. our sensys experiments were actually conducted in a > small area, with multiple hops enforced in software. so even if a node has > 60 neighbors, FTSP works fine. > brano _______________________________________________ Tinyos-help mailing list Tinyos-help@millennium.berkeley.edu https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help