Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Max White]

Ok, so realisticaly taking the less then legal approach isn't the easist corse of action.
Next up is if this came to some level of pesonal expense to somebody.

[Virex]

Well perhaps if we implemented some sort of 'smart network' rather then random and brutal transfers. So, for example, because we are transfering bits for free now, when your not using your bandwidth for anything intensive, your computer would volenteer itself to act as a server and take care of all the busyer computers around it.

The problem isn't too little nodes, quite the opposite actually. When one node is sending to another node, other nodes that can reach the second node can't send on the same frequency or else the messages will interfere and the second node gets nothing but garbage (known as the Hidden node and Exposed terminal problems). This is usually solved by using small "request to send" messages to essentially desynchronize the local part of the network, but this means that if you're trying to download something, you've got to share connection time with everyone else within your radius that's also communicating with the network. 10 persons on a 20 mbit network would only get 2 mbit if they are connecting at the same time and that's ignoring the additional overhead caused by network maintenance, request to send messages and actually finding your resources.


This problem could of course be reduced if you've got a stable mesh, because then you can run a link state routing protocol like OLSR (with some modifications to account for link quality and probably also Fisheye routing to account for scalability problems) with comparatively little overhead (you can expect nodes to stay put and run for an extended amount of time), but it'll be hard to get rid of collisions. Perhaps a system of frequency distribution could work, but that'd add extra overhead to the network and make bootstrapping more difficult to boot. My proffessor on this subject was quite clear on this: Wireless networking is unlikely to replace infrastructure-based networks. Instead it'll extend those networks to places where infrastructure is inoperable or too expensive to place.


The idea behind that is essentially that if you can build a mesh from where you are to an access point, you're set. In a city for example, access points could be placed at specific points with intensive traffic and if enough people enable their telephone as a node then anyone within the city center can easily reach the access point, even if that access point itself only has a limited radius. Similarly, in a disaster area, such as after a flood or earthquake, one'd only need to set up one access point to for example satellite internet (or find a cable that's still working) and handle the rest via mobile nodes such as cars equipped with communication equipment. Communicating problems like collisions or traffic jams down a highway is also an example. In that case you only realy need to transfer messages if things are crowded, so if enough cars have communication equipment you don't even need an infrastructure.