Planning meeting

Whiteboard Photo

How often do we want to poll the sensors?
What did we decide on a method of acking?
- No, but two options. First, just acks when sink has received all data it is waiting for, then acks with which data it has received. Second is to ack more regularly which allows nodes to resend their data easier rather than waiting 'til right to the end --Jt14v07 15:24, 14 November 2010 (GMT)
The idea of having the id set by number of hops from config won't work - it's only valid in a straight line. If it's a mesh layout, there will be multiple nodes with the same id.
- It also won't work because the ID of the node (network wise) cannot be changed after it has been set. Don't know if we need some sort of routing mechanism or just broadcast (and filter) in the right direction --Jt14v07 15:24, 14 November 2010 (GMT)
We need to decide the method of networking, whether we want to create links or just broadcast. Making links would be the much nicer solution but may be more difficult to create an ad-hocy style network

For those editing code, it might be an idea to start putting comments at the top of each file describing what it does. I'm already confused about anything with hardware calls!
- Are we going to start a new area with the code we are actually going to use rather than keeping everything quite fragmented in different examples --Jt14v07 15:24, 14 November 2010 (GMT)
Are we using direct memory addresses to store data, or is there another routine to store stuff?
Ring buffer can be implemented with 2 pointers. I remember it in DSA in first year.

Devices never store data from other nodes, they store their own data until they have confirmation that it has been received by the sink
We want to reduce the number of transmissions the nodes closest to the sink has to make by waiting for data from other nodes and sending it all at once