Conference Paper

An Online Adaptive Model for Location Prediction.

DOI: 10.1007/978-3-642-11482-3_5 Conference: Autonomic Computing and Communications Systems, Third International ICST Conference, Autonomics 2009, Limassol, Cyprus, September 9-11, 2009, Revised Selected Papers
Source: DBLP


Context-awareness is viewed as one of the most important aspects in the emerging pervasive computing paradigm. Mobile context-aware
applications are required to sense and react to changing environment conditions. Such applications, usually, need to recognize,
classify and predict context in order to act efficiently, beforehand, for the benefit of the user. In this paper, we propose
a mobility prediction model, which deals with context representation and location prediction of moving users. Machine Learning
(ML) techniques are used for trajectory classification. Spatial and temporal on-line clustering is adopted. We rely on Adaptive
Resonance Theory (ART) for location prediction. Location prediction is treated as a context classification problem. We introduce
a novel classifier that applies a Hausdorff-like distance over the extracted trajectories handling location prediction. Since
our approach is time-sensitive, the Hausdorff distance is considered more advantageous than a simple Euclidean norm. A learning
method is presented and evaluated. We compare ART with Offline kMeans and Online kMeans algorithms. Our findings are very promising for the use of the proposed model in mobile context aware applications.

Download full-text


Available from: Theodoros Anagnostopoulos
  • Source
    • "In addition, we perform a comparative analysis of the proposed LP with popular prediction algorithms reported in the literature. Secondly, we propose a novel adaptive mobility prediction algorithm, which deals with location context representation and trajectory prediction of moving users [6], [7]. In this context ML provides algorithms for learning a system to:  clusters the user movements,  identifies changes in the user movements  adapts its knowledge structure to such changes, and,  predicts the future user location. "

    Full-text · Chapter · Jun 2013
  • Source
    • "Focusing on the concrete type of pattern learning and prediction calculation methods, the options are varied, starting from the use of well-known mobility models [6,22] to machine learning methods, both supervised (Bayesian approaches [23,24], neural networks [23,25], Hidden Markov models [26]) and unsupervised (clustering techniques [27], Self-Organazing Maps [24], Adaptive Resonance Theory [28]), or information theory techniques (Markov models [13,14,29], compression algorithms [7–9,15]) among many others. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Predicting users' next location allows to anticipate their future context, thus providing additional time to be ready for that context and react consequently. This work is focused on a set of LZ-based algorithms (LZ, LeZi Update and Active LeZi) capable of learning mobility patterns and estimating the next location with low resource needs, which makes it possible to execute them on mobile devices. The original algorithms have been divided into two phases, thus being possible to mix them and check which combination is the best one to obtain better prediction accuracy or lower resource consumption. To make such comparisons, a set of GSM-based mobility traces of 95 different users is considered. Finally, a prototype for mobile devices that integrates the predictors in a public transportation recommender system is described in order to show an example of how to take advantage of location prediction in an ubiquitous computing environment.
    Full-text · Article · Dec 2012 · Sensors
  • Source
    • "The drawback of this model is that it has significant storage requirements in order to store the user patterns. In addition, the model in [10] responds slowly to changes, thus, cannot achieve fast adaptation to previously unseen mobility behavior. "
    [Show abstract] [Hide abstract]
    ABSTRACT: We focus on the proactivity feature of mobile applications. We propose a short-memory adaptive location predictor that realizes mobility prediction in the absence of extensive historical mobility information. Our predictor is based on a local linear regression model, while its adaptation capability is achieved through a fuzzy controller. Such fuzzy controller capitalizes on an appropriate size of historical mobility information in order to minimize the location prediction error and provide fast adaptation to any detected movement change. Our prediction experiments, performed with real GPS data, show the predictability and adaptability of the proposed location predictor.
    Full-text · Article · May 2011 · Computer Communications
Show more