antenna-theory.com

[shannon]

Hi,
I have inquiry,
In case, if I have cluster of nodes:
The clusterization procedure involves the following steps:

1) Every node broadcasts its ID to its one-hop neighbors. So every node knows the number of its neighbors and their IDs.
2) Every node broadcasts the number of its neighbors to the adjacent nodes.
3) The clusterhead is the node, which has the maximum number of neighbors among the adjacent nodes. If the number of neighbors is similar in several adjacent nodes, the node with the minimum ID will be the clusterhead. The presented algorithm provides that the maximum route distance between nodes in each
cluster is two hops. So the distance between each node in a cluster and clusterhead is one hop.

I consider cluster to be a virtual antenna array with a non-regular structure. The distances between the nodes in the clusters are bigger than in typical antenna arrays, much bigger than the wavelength. These aspects make synchronization inside the clusters a rather complicated problem.

What is a method for evaluating the distance between nodes in this case?

[Schubert]

Use a tape measure

[shannon]

[Bobby Digital]

[shannon]

[wolberine]

are you looking for some automated way?

you could try synchronization through timing - similar to how a GPS system works. (or even USE GPS on each device).

each node could time how long it takes to transmit to the other node.

depending on how far away the nodes are and your initial location assumptions, you will have varying degrees of accuracy for a distance magnitude, but a distance vector might be harder to figure out.

[bigSteve]

"These nodes are assumed to be scattered from an airplane or helicopter"

I think this statement violates the original first statement:
"Every node broadcasts its ID to its one-hop neighbors. So every node knows the number of its neighbors and their IDs"

If they are thrown out somewhat randomly, the devices cannot themselves figure out who their one hop neighbors are without already knowing the structure.

I think you'll need a system whereby nodes sequentially assume the roll of clusterhead and attempt to process the data from the other nodes. Possibly adding acoustic sensors will allow you to use time-of-arrival delay to estimate position. The problem sounds exquisitely complicated.