Recently, I’m working on a scheduling algorithm on wireless sensor network with Matlab. But when it comes to the very first topology generation, it becomes complex than I thought.
Here is what I have done:
- random generate nodes on planar with uniform distribution, remove nodes that’s too near to existing nodes.
- generate mesh grid topology with a coverage range, nodes within the range can establish connection.
- set the geometry middle point as root,generate tree topology by Dijkstra’s algorithm.
And here is preliminary result and codes.
The result is too simple and uncontrollable now. Most importantly the generated tree seems imbalanced, i.e the degree/number of children vary much, which seems not expected. Obviously, I need more constraint to make it more realistic.
I don’t need to carefully place the nodes to achieve minimum power consumption, nor to design channel model or structure of nodes, but some baselines like best coverage, less nodes, less interference, less traffic load(in bit*meter) is expected.
I survey and find tool like NPART https://www.informatik.hu-berlin.de/de/Members/milic/NPART, and someone says maybe NS3 can help. But I have limited time on it, anyway it’s just part of my research.
Could anybody give me some easy-implement hints or useful tools?
here is my simple code
https://www.informatik.hu-berlin.de/de/Members/milic/NPART
A realistic topology relates somewhat to the application you wish to use the sensor network for. For field area monitoring your used uniform distribution could work as realistic enough. On the other hand, depending on the deployment type there may be no need to remove nodes that are close by and you may find disconnected nodes not in reach of other nodes. Further on, a field sensor network monitoring e.g. soil would be lined up in rows adjacent to one another with node and row spacing matching your criteria. Another commonly used WSN topology is the Manhattan grid model, where nodes only occupy the "streets" of a urban grid topology. There nodes are randomly placed within those "streets".
Of course, as you wish to utilise a wireless network, the radio channel (or optical) comes into play. I don't know how deeply you are planning to take the channel into account in affecting propagation loss, fading, communications range, bit error rates, applied modulation and coding, obstacles, etc. If you are planning to take some of these into account, then the factors should match your application space. Attached is a publication with so basics on the RF propagation part.
Article Multihop Medium Access Control for WSNs: An Energy Analysis Model
- Badges
- Science topic