The main goal of the project is the development of methodologies, techniques, and tools for the management and the analysis of large amounts of data obtained from a spatio-temporal tracking of moving objects, basically only knowing their spatial positions at some specific time points. The term object refers to the generality of possible mobile devices that can be monitored, e.g., they can be associated with vehicles, animals, or people.

 

 

 

 

 

Trajectory extraction. The first part of the project will be devoted to the analysis of the problem under consideration and to the evaluation of available solutions for the management of data obtained from position tracking of moving objects (Mobile Object Databases). In addition to concerns about the filtering of the raw data as obtained from the position sensors, we will focus our analysis on issues related to the compression of the generated data stream. Then, we will address the problem of  applying suitable interpolation techniques in order to obtain a good approximation of the actual movement of the objects on the basis of the detected positions. In many situations, these techniques allow one to successfully reconstruct the most likely trajectory followed by the object. Next, the analysis will focus on the various proposals in the literature for an efficient storage of the trajectories. Most of them apply suitable segmentation policies, that partition a trajectory into a number of sub-trajectories. Each sub-trajectory is characterized by the fulfillment of specific semantic properties (for example, a trajectory can be described as a sequence of phases of stillness and movement). This analysis will identify the most-suited models, data structures and algorithms for the representation and the analysis of the trajectories followed by a specific object over time; it will also allow us to identify sets of objects that exhibit a similar behavior (same trajectory and/or same sub-trajectories).


Trajectory analysis. The second part of the project will focus on the study of the tools available for the analysis of the trajectories followed by the moving objects, ranging from data mining tools to statistical analysis techniques. Relying on the assumption that objects are not moving randomly, but, on the contrary, exhibit regularities / recurrent behaviors over time (partially induced by the underlying infrastructure), tools to extract high-level information from sequences of raw data will be identified and/or developed. A central role will be played by (temporal) data mining. Data mining aims at exploring a large set of data in order to (i) identify regularities, (ii) extract high-level information (semantics) on the behavior of the analyzed objects (places where the objects stop for a considerable amount of time, frequent trajectoriesm and relationships between time periods and corresponding movements), and (iii) formulate general rules encoding the behavior of objects. A special attention will be given to techniques and tools for the aggregation of data from different trajectories and to the analysis of the resulting aggregations. As an example, the sequence of positions over certain time periods characterize the behavior of the considered object/objects and can be further analyzed to interpret and contextualize a phenomenon, to identify recurrent patterns, or to predict future trends. Some examples of high-level information that can be extracted from trajectories are (i) areas in which an object usually spends a lot of time, (ii) paths that are repeated on periodical basis, and (iii) paths common to sets of objects. Obviously, in all these cases a fundamental role is played by the time period in which the phenomenon occurs (morning, working time, night, ..)


Prototype realization. On the basis of the outcomes of the first two phases of the project, we will outline the software architecture of a system for the management and the analysis of data about a set of moving objects, and we will implement a prototypical version of such a system for an experimental evaluation on real data sets (relative to the domain of positioning systems). Besides a data warehouse component, the system will offer a set of basic and advanced functionalities for the analysis of the behavior of moving objects. More precisely, the system will feature the following three fundamental components: (i) a data warehouse, constantly feeded (possibly at a high rate) with data about the positions of moving objects; (ii) a set of algorithms for compression, filtering, segmentation, data mining, and data analysis; (iii) a user interface that allows a non-expert user to evaluate the semantic information that the system is able to extract from the raw input data stream.