[net2-wg] Packet interface

Philip Levis pal at cs.stanford.edu
Wed Nov 30 13:23:04 PST 2005


On Nov 30, 2005, at 11:06 AM, Henri DF wrote:

> I don't quite understand this. Even if each SendPoolEntry has a fixed
> un/urgent bit, the component that provides the SendPoolEntry's can  
> still get an
> un-urgent packet from one SendPoolEntry user, followed by an urgent  
> one
> from another.

Yes. Assume a low-power-listening CSMA data link. A network protocol  
starts its entry with the urgent bit set to false. The pool  
implementation holds onto the entry for a while, waiting for an  
opportunity to piggyback this packet on other traffic to the  
destination (possibly from another node). At some point, it will bite  
the bullet and send a long preamble to wake up the other node, but  
it'll wait a bit first. If, while it's waiting, another network  
protocol starts its pool entry with the urgent bit set to true, then  
the pool will send the urgent packet with a long preamble and can  
piggyback the un-urgent packet.

If the urgent and not-urgent packets are to different destinations,  
then it's up to the implementation whether it needs to send multiple  
long packets. I can imagine using both approaches. A node that hears  
a long preamble might stay awake as long as it hears packets for  
itself or anyone else, or it might shut off as soon as it hears a  
packet for someone else. There's an RX/TX energy cost tradeoff there  
whose decision should be up to the implementer.


> What about destination, does a routing protocol have to allocate  
> one pool
> entry per possible destination?

This interface assumes that a protocol has a fixed number of pool  
entries. If it has more destinations than pool entries (the basic  
case being one pool entry), then it needs to figure out how to  
allocate the pool entries to destinations. It could do something like  
this:

configure entry for destination A
send 4 packets to A
configure entry for destination B
send 2 packets to B

there's a tradeoff here between the amount of communication  
parallelism and the state you allocate (pool entries).

Phil

-------

"We shall not cease from exploration
And the end of all our exploring
Will be to arrive where we started
And know the place for the first time."

- T. S. Eliot,  'Little Gidding'





More information about the net2-wg mailing list