Plotting coordinate data being generated in a separate thread

t.vldz

Plotting coordinate data being generated in a separate thread

in Contributed Library Questions • 1 year ago

Greetings,

I am trying to display (many) points on a map from a stream of coordinate data. Specifically, I want to display the locations of IP addresses on a world map. I thought I had the hard part out of the way, and I usually figure these things out by myself, but I'm having some trouble finding a way to actually display the data that I am generating. I hope that this is a general design problem, and not some limitation. Some background:

I'm using the jflow library to capture IP addresses from a router in real-time.
I'm using the GeoIP library to generate long/lat coordinates from these IPs.
I plan on using the Unfolding library to display maps, and draw points.
I'm using Processing (through Eclipse) to draw the output.

I am able to capture and print IP addresses and coordinates to the console. However, I cannot find a way to actually USE this data within the Processing draw() method. I think that this is because the jflow library collects (and does whatever else you want) in a separate accounting thread. " Note that the collector is multithreaded. Calling start will start collecting in the background, the call will return immediately." If this is the case, what's the best way to access and draw the incoming data? Some kind of IPC? Somehow make the IP collection thread use the draw() method?

Below is an example IP collection class, with some edits showing how I would like to use it. What can I do with this separate thread?

This might lead to other questions, like, should I somehow run the draw() method at a constant rate and asynchronously update it with incoming data from the IP collection thread? Any advice is welcome. Please let me know if anything is not clear.

import java.net.InetAddress;
import java.net.SocketException;
import java.net.UnknownHostException;
import nettrack.net.netflow.*;
import com.maxmind.geoip.*;
import com.maxmind.geoip.Location;
public class V5Printer
{
static class PrintAccountant
implements Accountant
{
public void account(Flow f)
{
//method nest below are not real, just illustrating intent.
drawMapCoordinatesPoint(getGeoIPCoordinates(f.ipAddress))
}
public void destroy()
{
}
}
public static void main(String[] arg)
{
InetAddress source = InetAddress.getByName("10.10.10.10");
Collector collector = new Collector(2055);
V5FlowHandler handler = new V5FlowHandler(source, 100);
handler.addAccountant(new PrintAccountant());
collector.addFlowHandler(handler);
collector.start(); //starts the collection thread.
}
}

Replies(3)

PhiLho

Re: Plotting coordinate data being generated in a separate thread

1 year ago

Moved to this section because your code depends on several of these libraries, and the answer might rely on knowledge of these libraries.

I think the collected data should go in a collection (eg. an array list or a linked list), whose access (add item, access the collection) should be synchronized (to avoid update while reading). Although a linked list might be safe (adding only at the end without change of the remainder).
Then draw(), which is automatically called at regular intervals (60 times per second by default, if it is fast enough) can walk this collection and draw the items.
Thus, the asynchronous updates should not interfere with the drawing process.
To test carefully as concurrent programming is delicate to get right...

t.vldz

Re: Plotting coordinate data being generated in a separate thread

1 year ago

Thanks for the input. I guess I would need to use something from the Java concurrent library? Something like this?

http://docs.oracle.com/javase/1.5.0/docs/api/java/util/concurrent/ConcurrentLinkedQueue.html

Since I'll be dealing with a stream of infinite data, should I just process X amount of items in the queue during each loop of draw()? Is there any other advice that you can give related to drawing from a constant stream of data?

Thanks.

PhiLho

Re: Plotting coordinate data being generated in a separate thread

1 year ago

Indeed a Queue, working as a FIFO, might be well suited to your needs: thus you can fill the queue at one end, and empty it at the other end.
I actually don't know if you need something handling concurrency. Perhaps you can try with a simple structure (probably faster) and change to a concurrent queue if you have issues.
I don't have an infinite stream of data at hand to test the algorithm...

Top Reply