||Wireless Mesh Networking|
|Subject:||Ad-hoc mesh routing is not a silver bullet...|
Response to: Ad-hoc mesh routing is not a silver bullet...
The Grid people from MIT have deployed an ad-hoc mesh routing wireless test network called Roofnet (http://www.pdos.lcs.mit.edu/roofnet/) which seems to work for them quite nicely.
To me, their biggest improvement is choosing routes not only based on hop counts and binary availability of nodes (either "available" or "unavailable"), but on link quality, stability and retransmission ratio. This takes the WLAN-specific link characteristcs into account, which differ a lot from regular wired networks where links are either up or down and very less likely flaky.
One advantage of the Mitre MobileMesh is that it is pretty easy to set for a small test mesh, while I agree that the MIT Grid has solved problems in this areas others are still working on. Whatever implementation is preferable, wireless mesh networking seems to have a great potential in certain areas.
Mesh you again on the next Freifunk Summer Convention in Djursland! :)