Hi Paolo, I'm only into the Boomerang (Moteiv) version of MultiHopLQI, but the end-to-end link quality already consists the distance in a way. The LQI is represented in costs (good LQI -> low cost, bad LQI -> high cost), which will be summed up from every node of the path. So if a path has many hops the costs will also be high. Off course it is possible that a path with short distance (few hops) and bad LQI values causes higher costs than a path with more hops but better link quality. But saving all partitions of a path in a history would need a lot of memory.
Also the LQI / RSSI values should be a representative of the distance according to the free space model of Frii. And describing the distance with RSSI is more precise than hop counts (but not more accurate). But you can easily add the parameter hopcount to the parent table and do some other changes in the code so that in the parent selection the costs will only be considered if the hopcount is equal. Regards, OLE -----Ursprüngliche Nachricht----- Von: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Im Auftrag von Paolo Masci Gesendet: Montag, 10. September 2007 14:40 An: [email protected] Betreff: [Tinyos-help] Cycle detection in multihoplqi (surge) for tinyos-1.x Hi, I was giving a look to the multihop routing algorithm implemented in tos/lib/MultiHopLQI. I have noticed that nodes select their parent evaluating the end-to-end link quality of the path to the base station. In order to avoid cycles, shouldn't they select their parent evaluating also the distance from the base station? Without the check on the distance, there is no guarantee that the packet gets closer to the base station: isn't there the risk of cycles? (for example, I guess it may happen in scenarios with long paths and with link quality that frequently changes) The developers of this library, besides checking the address of the node that originates the packet, included also a check on the history of packets seen by the node, but this guarantees no cycles only for paths whose length is less or equal to the history size. Did anyone else notice this? Am I missing something? Regards, Paolo _______________________________________________ Tinyos-help mailing list [email protected] https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help _______________________________________________ Tinyos-help mailing list [email protected] https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
