[Tinyos-host-mote-wg] Re: [Tinyos-2.0wg] Bridge, agenda for TinyOS 2.x 10/5

[Kevin Klues]

David Gay wrote:

>There's two obvious reasons why this isn't that appealing:
>- The B-MAC LPL scheduling code isn't complex enough to be worth
>separating (there's not much complexity in "every N ms" ;-)). And
>where it is more complex, it's tied to sending or receiving states
>(e.g., if no message is received within X ms of apparent channel
>activity, go back to sleep).
>- the LPL code needs to know the radio's sending/receiving/etc state -
>LPL-checks are only performed if the radio is idle.
>
>To summarise, the interactions between LPL, carrier sensing, preamble
>detection and noise floor estimation are complex enough that it's not
>clear that trying to break them apart is fruitful.
>
>David
>  
>
I understand the desire to keep them integrated if the only MAC 
impelemented is going to be B-MAC.  The problem arises when trying to 
use TinyOS to evaluate other MAC protocols that have not been 
implemented yet.  This seems to me like a reasonable separation to make 
if it would ease the task implementing MAC protocols that use a 
different sleep scheduling algorithm other than LPL, but are independent 
of the CSMA/preabmbel detection/noise floor estimation schemes used.  
I'm not saying it will be easy to do the separation and keep all of the 
timing/efficiency/code size semantics the same, but I DO think its worth 
the effort to investigate whether it is possible or not.

Kevin
_______________________________________________
Tinyos-2.0wg mailing list
Tinyos-2.0wg at Mail.Millennium.Berkeley.EDU
https://mail.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-2.0wg