[SensorNetArch] MAC + Network Layer Idea

Philip Levis pal at eecs.berkeley.edu
Wed Oct 27 10:33:07 PDT 2004


On Oct 25, 2004, at 11:06 AM, Cheng Tien Ee wrote:

> Hi,
>
> The paper on Synopsis Diffusion has the interesting idea in which 
> multiple parent motes receive and aggregate data. It is more efficient 
> if the child mote broadcasts once, and its parents listen to and 
> receive that one packet. Currently, the packet header specifies that 
> the packet is either destined for a particular mote, or should be 
> received by every mote in range (broadcast). At this time (I believe) 
> it is not possible to specify a subset of receivers, in this case, the 
> parents. The definition of who the parents are is dependent on the 
> sink mote, that's to say it's dependent on information at the routing 
> layer. Maybe it's a good thing to define a TOS_PARENT_ADDR: when a 
> mote's MAC layer receives that packet, it passes it up to the routing 
> layer to decide if the packet should be accepted (or perhaps that 
> information can be passed to the MAC layer). The network layer needs 
> to check that the mote is indeed the child mote for that particular 
> destination (i.e. extra state needs to be stored).
>
> Another application would be that of routing along multiple paths 
> (more resilient to any one path failure, brought up by Rodrigo).

I know Sam explored something like this in TinyDB. I don't know if he 
ever implemented it beyond simulation, however. There was also a paper 
in HotNets by Robert Morris about multiple packet destinations, in this 
case for redundancy and reducing hops.

It's important to note that destination address can be separate from 
nodes that "receive" the packet. The question is merely where in the 
stack a node filters packets that are not addressed to it. This can be 
in hardware (e.g., the CC2420), which limits the options, but certainly 
stacks like GenericCommPromiscuous, or the Snoop handler in multihop 
routing, allow a node to overhear other node's packets. Without the 
latter, epoch synchronization in TinyDB would become much more 
difficult.

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 SensorNetArch mailing list