Presentation on theme: "Say NO to Multi-hop Wireless Networks Suresh Singh."— Presentation transcript:
Say NO to Multi-hop Wireless Networks Suresh Singh
It is a matter of definition General research in the area appears to be fairly irrelevant –Lots of really clever ideas have been developed in: MAC, Directional or Omni, Smart and other types of antennas Routing protocols TCP over multi-hop Low-energy considerations Capacity Etc., etc., etc.
How much of this has been used? Not a whole lot Why not? –Most of the time, we start with an abstract model along the following lines: nodes are randomly distributed, some may be mobile and they need to form a network –This type of model does not appear much in practice and hence the developed solutions are not useful in real scenarios
Some problems with using the typical abstraction PHY layer is very important - generally abstracted away –For instance, propagation is highly complex and can make or break a system Cooperation between participants is always assumed (though there is some incentive based work) –If Alice is a low-volume user, while Bob is a high- volume user, Alice may need to sacrifice her power for Bob. Why would Alice do that?
Consider examples Mesh networks: Are deployed, but have specific properties (point-to-point links) that are not part of the model of most multi-hop work. VANETs: Within the vehicle, a single cell architecture makes sense. Between vehicles, the communication model is different from typical multi-hop networks. Sensor Networks: Maybe I should keep quiet….
Cart before the horse Problem appears to be that we started from an abstract definition, solved it, then start looking for solutions. This is ass backwards. We should solve real problems directly. If common principles then emerge that is great. But that should not be the goal. Look at other fields -- Biology or Electrical Engineering. Flavor of work is completely problem driven.
Solutions not rooted in real problems are doomed to obsolescence