Gaggle
Gaggle copied to clipboard
geeksville
UDP livetrack24 protocol
the Leonardo Live24 communication over UDP has been implemented or not yet?
Email from Manolis
Here is the info the UDP/TPC session aware protocol : (i am using the word midlet since my java app is a midlet)
the IP to use for testing is 188.40.246.133 port 995 UDP (both UDP / TCP ) and the tracks are displayed at test.livetrack24.com
in the zip file are the relevant classes and a sample code to send the packets is this
We can login into the server with user/name password. Logged in users have more functionality like triggering their predefined SMS alerts and more stuff. an account is not mandatory though , the user has 2 options: a) create an account and then login once from midlet. his userID will be stored in the middlet and the sessionID that is send with each gps packet has his userID embedded in it. thus even if the first live packet that has login information is lost, the server still knows the userID of the packets. b) dont use an account. the sessionID is a random 4byte long and the user/password pair is send with the first packet to the server. all other packets are linked to the user via the unique random sessionID. if the first packet is lost, the user is displayed as guest. the recommended way to send the first packet is this: - use TCP instead. - expect OK as an answer. if not retry at predefined intervals. not too many / not too frequent : TCP can be much more verbose, a TCP packet can be send back and forth many times before it timeouts, so a simple 100 bytes TCP packet can generate even 1-2 kb of GPRS traffic. here is my intervals:
// send login intervals in seconds:
// 2 min,2 min,5 min,10 min,30 min,30 min,30 min
// total 6 retries ( the first one is not counted )
int[] sendLoginIntervals={120,120,300,600,1800,1800,1800};
I'll EXPLAIN METHOD B) , but A) is similar, you only set the sessionID in a format that the server knows it is a registered user
HERE IS a workflow:
Sending the first (login) packet GpsTrackLive.sNet is an static instance of LiveNet, the first thing that does in the run() is send the login.
GpsTrackLive.sNet.open();
You app must monitor the success of the sending of the logjn packet with some code and retry like that:
if (!GpsTrackLive.sNet.loginHasBeenSent) {
// send login intervals :
// 2 min,2 min,5 min,10 min,30 min,30 min,30 min
// total 6 retries ( the first one is not counted )
int[] sendLoginIntervals={120,120,300,600,1800,1800,1800};
//int[] sendLoginIntervals={20,10,300,600,1800,1800,1800};
// now is a good change to resend login info...
if (GpsTrackLive.sNet.sendLoginTimes < sendLoginIntervals.length ) {
//GpsTrackLive.logMsg(MIDPLogger.INFO,"sendLoginTimes: "+GpsTrackLive.sNet.sendLoginTimes
// + " resendLoginTimer: " +resendLoginTimer);
if ( resendLoginTimer > sendLoginIntervals[GpsTrackLive.sNet.sendLoginTimes]* 1000 ) {
GpsTrackLive.logMsg(MIDPLogger.INFO,"Will resend Login information #"+GpsTrackLive.sNet.sendLoginTimes);
resendLoginTimer =0;
GpsTrackLive.sNet.sendLogin();
}
}
}
Now on to gps point gathering and sending....
in the normal GPS/ TRACK loop we first STORE n gps points (n=1-5) with this :
short cog;
byte sog;
byte[] netBytes=null;
try {
sog=(byte)gps.getSpeed().toLong();
cog=(short)gps.getHeading().toLong();
netBytes=PrepareMsg.prepareMessage(tm,lon,lat,alt, sog, cog);
} catch ( Exception ex01) {
GpsTrackLive.logMsg(MIDPLogger.ERROR,"Track.run: Exception 01:" + ex01.getMessage());
}
try {
// store this in the buffer
if (netBytes!=null){
netStore[netPointsStoredTmp]=netBytes;
netPointsStoredTmp++;
netPointsStored++;
GpsTrackLive.logMsg(MIDPLogger.DEBUG,"Track.run: netPointsStoredTmp: "+netPointsStoredTmp);
} else {
GpsTrackLive.logMsg(MIDPLogger.ERROR,"Track.run: netBytes is null ");
continue;
}
} catch ( Exception ex02) {
GpsTrackLive.logMsg(MIDPLogger.ERROR,"Track.run: Exception 02:" + ex02.getMessage());
}
when we gather the mSendEveryN packets it is time to send them .... SPECIAL CASE: if it is the first point , we send it right away, we want the position of the user to be available on the web as soon as possible.
// now decide if we store this record or send it
// take special care for the first gps fix : send right way
if ( netPointsStoredTmp == mSendEveryN || packetsTransmitted==0 ) {
// the time has come, send the buffer and reset recordsStored
/*
header 1st byte DEFAULT VALUE of GPS points
1 1=gps 0= info 1 // GPS point
2 1=request 0=signal 0 // signal
3 binary=1 text=0 1 // binary
4 1=simple header 0 = extended 0 // extended
5 1= no ack required 0=> send ack +status 1 // no ack
6-8 0-7 gps packets num ( +1) 3 // means there are (3+1)= 4 gps points in this packet
so is 1-8 packets num // or type of text 1->login
Next 4 bytes: sessionID
Next 2 bytes: packetID
So the header of each packet is 1 + 4 + 2 = 7 bytes
the gps data for each point is 16 bytes
# 4 4 4 2 1 1
# tm lat lon alt sog cog
*/
// isGPSdata,isRequest,isBinary,isSimpleHeader,noAck,packetNum
byte[] headerBytes=LiveProtocol.preparePacketHeader((byte)1,(byte) 0,(byte)1,(byte) 0 ,(byte) 1,(byte)(netPointsStoredTmp-1) );
byte[] msgBytes=PrepareMsg.concat(headerBytes,netStore,netPointsStoredTmp);
GpsTrackLive.logMsg(MIDPLogger.DEBUG,"Track.run: Send packet # "+packetsTransmitted );
// log the packetID and time
preferences.lastNotTerminatedSessionID=LiveProtocol.sessionID;
preferences.lastPacketID=LiveProtocol.packetID;
preferences.lastPacketTM=(int)(System.currentTimeMillis()/1000);
preferences.saveRunning();
latestNetBytes=msgBytes;
if ( GpsTrackLive.sNet.sendNetMsg(msgBytes) ) {
packetsTransmitted++;
} else {
packetsError++;
}
netPointsStoredTmp=0;
}
after the track finishes we need to also let the server know that and also ask the user if he had any problems... (good for extreme sports etc....)
see SendEndOfTrack class for the code that does that....
Hi chanoch,
And thanks for your interest in gaggle !
I'd also like to see the (new?) UDP protocol of livetrack24 implemented in gaggle, but the classes you are showing might not be compatible with the current gaggle licences for direct use it says : "Copyright (C) 2006 Dana Peters"
I thinks that if you plan to re-code and only take that as reference, it might be OK, but if you mean to use them "as is" or even after modification, we probably need to know what licence it is.
I didn't see that one first, sorry. I had nearly the same comment as sylvain in the PR :).
In principle, you should not even read this code before re-coding your own, as it can be seen as "derived work"...
