2
$\begingroup$

This question is a part of an online course (Mobility Data Management) that I am currently auditing. A part of the project is to implement a system that can predict user's future location. This kind of application has recently got some interests from database and mobile application developers.

This is the problem statement: Assume that we are given a road network G (V,E), the recent motion history (trajectory) T, and current instant velocity v of a user, a temporal duration dt, and (optional) a historical trajectory database D of other users’ movements. Our goal is to estimate user’s location after dt.

I found that the following paper has a good description about what a road network and the trajectory data is: http://www.sis.pitt.edu/~xliu/papers/ceus.pdf

Does anyone has an idea on how to formulate a simple solution of this mobility awareness problem (predicting user's future location), and how to start the implementation (how you formulate the problem in an OOP way, such as: what classes are needed to represent the problem correctly). I cannot find the correct tags for this question, probably the moderator can add new tags such as: graph-networks, mobility-data, trajectory, etc.

$\endgroup$
1
  • $\begingroup$ The last part of the question about implementation is not theoretical computer science and is off-topic here, you should ask it on the SO, please read the FAQ. (ps: the first part seems more like a machine learning problem suitable, you may want to check metaoptimize.) $\endgroup$ – Kaveh Jul 17 '11 at 6:46
1
$\begingroup$

My professor uses a hidden markov model based upon the previous routes that the bus has been. See http://www.cs.uic.edu/~jakob/papers/transitgenie-sensys09-demo.pdf for a paper on it. Although the paper doesn't go into the architecture I remember it from a lecture he gave on the application.

$\endgroup$
0
$\begingroup$

Maybe you should have a look at this article.

Spatial regression is applied to GPS floating car measurements to build a predictive model of road system speed as a function of link type, time period, and spatial structure.

$\endgroup$
1
  • $\begingroup$ I just read this article. It has no relation to my problem above. This paper talks about predicting car speed based on the car data movement and other neighborhood factors. While my problem is more about predicting where the user's future location is based on his movement data history and other users' movement history. Thanks anyway ! $\endgroup$ – all_by_grace Jul 21 '11 at 14:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.