Bluetooth + system.getNano - distance?

Question

Hi, not sure exactly what forum to post this in so I'll take it here...
Anyway I recently read an article about some new patents by apple, one which involved bluetooth finding lost keys and such via "hot/cold" feedback (using one device to find the other) and the other using triangulation (multiple devices) to more exactly pinpoint the lost device. I haven't looked at their patent so not sure exactly what it entails but from what I could get it worked similar to a gps but much smaller scale (it said something about atomic synced time for instance).
Anyway the reason I found the article was because a colleague of mine was wondering if it was possible to create something like this.
Now I'm most familiar with java so if possible it would be nice if I could use it... however got some questions on how plausible it is. And just cause I'm most familiar with java don't mean I'm all that good at that either ^^;
Anyway, to make it simple lets skip the triangulation bit. We have 2 devices and we would like to find the distance between them. Now I read an article about using signal strength as well to do this but their conclusion was it didn't work very well. So lets skip that as well. And since I doubt I could sync two devices atomically and then as the GPS works send the time and calculate the distance from that.
So my idea was this, since the clocks are not synced, could you use the round trip time (RTT) /2 to calculate the distance?
pseudocode of the operations
client1
OpenConnectionToOtherDevice connection;
is = connection.getInputStream();
os = connection.getOutPutStream();
start = System.getNano();
os.write(1);
is.read();
rtt = System.getNano() - start;
distance = speedoflight * (rtt/2) / 10^9  // last division to get it to secondsclient2
waitforconnection();
open inputstream is, openoutputstream os;
is.read();
os.write(1);The code above would also to improve accuracy not be allowed to be swapped out and such...
Would something like that actually work? :S From what I've read as well the getNano uses the most accurate clock on the system. "On XP the best you can get (the windows performance timer) is microsecond timing (10^6)"
Which wouldn't give high enough accuracy to calculate a distance in metres. But lets assume we get it in nanoseconds... Would something like this work or would you get a lot of delays on other parts on the code? :X
Might not be entirely clear there towards the end... I'll try to make myself more clear if you got questions.
Feedback appreciated
Christian

Answer 1

j2ee_ubuntu:
Yeah, it is to get a RTT, since the clocks aren't synced you cant send the the timestamp when it was sent and then compare that to the internal clock on the receiving device.
db:
Yeah I read some of that and was kinda what I was afraid of... I get an average of just below 2000 something...and since like 100ns results in ca 30m, calculating a mean and substracting that is also out of the question obviously...
Do you know if it's like possible (I think it was in some assignment way back in school in assembler) where you force a section to run, and it can't be placed on the stack before it's done it's stuff... :X
I'm guessing you'd have to go lowlvl like this for it to work properly in the first place, or anyone else got any ideas? =)