On Tue, 9 Dec 2003, Anthony Townsend wrote: ]hmmm.... i can see a # of interesting problems right off the bat. ]For example - what about accumulated errors when you start ]daisy-chaining inferred locations?
Triangulation. At least where I move about in NYC there are never less than 3-4 active WiFi nodes about transmitting beacons. You don't need to connect to use them for location services. If you see 5 networks and 3 are in the database already with different locations then you already have some idea where the other two are and have more knowledge about the three in the database than you did before.. really what you build up is a sampling of the coverage of the nodes, something interesting in itself. Letting the user correct the map by saying, "um no I'm actually 30 feet from where you planted me on the map" gives you a sample, "at 42nd and Lex we see node 123 at -20dB, node 231 at -30dB etc..." ]for example. Node A is at 10th st and 3 rd ave, with a definiteive ]location in the database. Node B is discovered at 11th st and 3rd ave, ]but stilll in range of Node A, so geocode it to Node A proximity. Node ]C is discovered at 12th but still in range of Node B, so geocode Node C ]to Node B (which is already geocoded to Node A location) ]so now i'm on 12 th street, but the database has been updated to say ]i'm really on 10th street. So you correct it and the next time you sync the database gets better. You could bootstrap the database with the locations of all the Verizon AP's or all the nodes from one of those netstumbler created databases. But really nothing should have a definitive location, if the app is popular enough it will quickly become the most accurate map of nodes and their coverage possible, even cell phone services can only get location information from their own customers. Marking the free-beer nodes could be nycwireless' contribution, since this map, with nycw nodes marked, would be better than our current map for finding nodes to connect to on the run. Though our nodedb map would still be more useful when you wanted to create person to person links, with antennas modifying the "coverage". ]also, i'd like to see this broadeneed a generalized infrastucture for ]proximity-based automated positioning - incorporating any unlicensed ]wirelses technology - WiFi, Bluetooth, UWB. GPS won't be working well ]in Manhattan-like settings anytime soon, and the really interesting ]places are going to have these types of networks anyways so why not ]leverage them.... let the architecture speak! Heh, it would be really neat if this killer app was created and GPS suppliers started incorporating it in their products as a supplementary location service for cities. If I only had 40 hours in my day... -- Daniel << When truth is outlawed; only outlaws will tell the truth. >> - RLiegh -- NYCwireless - http://www.nycwireless.net/ Un/Subscribe: http://lists.nycwireless.net/mailman/listinfo/nycwireless/ Archives: http://lists.nycwireless.net/pipermail/nycwireless/
