[Tinyos-help] Custom MAC Implementation Question

Michael Schippling schip at santafe.edu
Tue Sep 27 17:50:02 PDT 2005


what's a "hidden terminal" problem? and other issues?
(not to question your architectural decisions, but WTF?)

I did a master-slave round-robin kinda thing with, heh,
Lego Mindstorms IR comm once, but that had the master to
provide sync. With an ad-hoc-ish network
how do you plan to synchronize your time slices?
token passing? what about missed message recovery?
and if you don't rip the guts out of the under-layers
you're still going to have send delays and the other
baggage of the current MAC....

MS


Brett Parsons wrote:
> Hi All,
> 
> I'm looking to implement my own "global" MAC protocol in order to 
> eliminate issues such as the "hidden terminal" problem.  I want to use a 
> very simple approach that involves splitting the send window into n 
> slots, where n is the number of nodes in the network.  Nodes would only 
> send when their slot comes around.  Although I know this is inefficient, 
> it will suffice for my research purposes.
> 
> My question relates to how to implement such a protocol.  Currently, a 
> node can send 3 types of messages in my network:
>    -AM_INTEREST
>    -AM_DATAREADING
>    -AM_REINFORCEMENT
> To accomodate this, I have three parameterized instances of SendMsg.
> What I would like is a layer that sits on top of "SendMsg" that will 
> only call SendMsg.send() when the node's time slice is active (i.e. - 
> set a timer, wait for x ms, then send).
> Does anyone have any experience with implementing anything similar to 
> this?  Any suggestions on the best software architecture approach 
> required to accomplish the task?
> 
> Thanks,
> 
> Brett
> 
> _______________________________________________
> Tinyos-help mailing list
> Tinyos-help at Millennium.Berkeley.EDU
> https://mail.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help


More information about the Tinyos-help mailing list