This is a technical paper about ad-hoc network implementations. I don't normally focus much on technical papers, because I think the most interesting challenges revolve around humans and how they use technology. But the technical issues do need to be solved. This paper discusses gossip-based multicasting, mesh networks, geocasting, and anycasting.Group Communications in Mobile Ad Hoc Networks
Prasant Mohapatra, Chao Gui, Jian Lipdf
It's possible that I'm not technical enough to critique this properly, but I get the impression that network design isn't user-centered. When engineers think up new protocols they think about reliability, bandwidth, efficiency and perhaps if we're lucky, highly theoretical and detached usage situations and devices.
The paper mentions: "In addition to performance, some common issues that researchers have considered while designing most group communication protocols include energy conservation, reliability, security, and QoS support."
It has been said that bluetooth was doomed to be difficult to use from day one because of it's lack of concern for usability when developing the initial protocol. I think it is perhaps time for networks designed around observed usage behavior. While network designers do need to think ahead and allow new behaviors to develop, a lot could be learned from modeling how different groups communicate now, using existing technology.
For instance: Will the users be mobile? If so, how frequently or for what duration? What geographical areas do they roam in? How do they want to exchange data? How close will they be to each other? How much are they willing to pay and are there ways to make data transfer cheaper? How much information do they want to share? How do various network characteristics affect sociability, and which are most suited to the group? What level of security is needed for different types of users or information? How long will they need to send information without recharging? What kind of infrastructure can be relied on to relay information in the users' environment?
Currently we seem to do a lot of "hacking" of protocols to get them do what we want. BitTorrent had to hack the technology to allow people to share efficiently by doing simultaneous upload and download. Wouldn't it be great if we had networks built to support how we naturally interact instead of just the requirements of information theory and physics?