DBLab School of Computer and Electrical Engineering KDBSL NTUA
Monday, March 30, 2020

1. General Framework 

Knowing where people are and how objects (e.g., vehicles) move in space is expected to be widely utilized in the near future. The so-called Location-Based Services (LBS, for short) take advantage of this basically spatiotemporal information so as to provide to users notification about events that just happened or are about to occur. Our approach stems from the necessity to manage and efficiently process the huge amounts of spatiotemporal and thematic data that are gathered in sources (e.g., sensors, GPS) and transferred through a network. Furthermore, developing capabilities to serve in almost real-time multiple concurrent requests from thousands of users is considered as a major challenge.

2. Research Issues

-Modeling and Representing Movement
T. Sellis, V. Kantere, K. Patroumpas, S. Skiadopoulos

We are working towards a model that captures the multidimensional nature of trajectories of moving objects, as well as the interaction between space and time, frequent positional updates and interrelationships with stationary entities. We will also attempt to formulate a query language adequate to express requests that could possibly be posed in LBS applications. Such a language should include spatiotemporal predicates, window specifications, as well as composite operators that involve spatial and temporal arguments.

We propose two architectural representations of data management for location-based services. In the centralized approach, information gathered by moving objects is forwarded to a central management system. Conversely, in the decentralized approach, the moving objects constitute a peer-to-peer network and become active members of the system: beyond gathering the information, they are responsible of processing it, too.

-Streams of Moving Objects' Trajectories
T. Sellis, K. Patroumpas


In a centralized data management system, information about objects' movement is considered to arrive as continuous, time-varying and possibly unbounded data streams. Several window constructs are being studied so as to extract from the original trajectories only a finite portion (usually the most recent tuples), exploiting their spatiotemporal variability when possible.

We are planning to devise optimization techniques for multiple continuous queries over trajectories. Although certain parameters may change dynamically, queries themselves remain active for a long time. Typical point, nearest neighbor, or range queries must provide their results incrementally, keeping in pace with fluctuations in the rate of incoming tuples.

Moreover, if approximate query answering is accepted as a trade-off in order to achieve real-time processing of requests, then probably the amount of data should be reduced. We aim to adjust several synopsis techniques (like wavelets, sampling or histograms) to the case of trajectories, taking into account the multidimensional nature of data related to movement.

-Managing Queries in Networks of Moving Peers
T. Sellis, V. Kantere


In the decentralized approach the goal is to study the viability of the dynamic processing dissemination in terms of both data and queries. In this context, the mobile objects form a peer-to-peer netowork and are responsible for the management of the spatiotemporal information and the propagation of it to other nodes of the system, whenever needed.

One of the main issues of the decentralized architectural model is the development of indexing and information dissemination techniques appropriate for the dynamic nature of a peer-to-peer environment. These techniques have to be adapted or, even more, exploit the special characteristics of the high-evolving spatiotemporal data of an LBS application. Furthermore, it is very important to propose routing methods for user requests for information, which are expected to be mainly continuous queries.

3. Selected Publications

V. Kantere and A. Tsois. Using ECA Rules to Implement Mobile Query Agents for Fast-Evolving Pure P2P Networks. In Proceedings of the International Conference on Autonomous Agents and Multiagent Systems (AAMS04), New York, USA, July 2004.

K. Patroumpas and T. Sellis. Managing Trajectories of Moving Objects as Data Streams. In Proceedings of the 2nd Workshop on Spatio-Temporal Database Management (STDBM’04), Toronto, Canada, August 2004.

K. Patroumpas and T. Sellis. Window Specification over Data Streams. In Proceedings of the IFIP Workshop on Semantics of a Networked World (ICSNW’06), Munich, Germany, March 2006.

M. Potamias, K. Patroumpas, and T. Sellis. Sampling Trajectory Streams with Spatiotemporal Criteria. In Proceedings of the 18th International Conference on Scientific and Statistical Database Management (SSDBM’06), Vienna, Austria, July 2006.

M. Potamias, K. Patroumpas, and T. Sellis. Amnesic Online Synopses for Moving Objects. In Proceedings of the 15th International Conference on Information and Knowledge Management (CIKM’06), Arlington, Virginia, USA, November 2006.

4. Contacts

Prof. Timos Sellis                      
Phone: +30-1-772-1601, Fax: +30-1-772-1442
e-mail: timos@dblab.ece.ntua.gr

Verena Kantere, Kostas Patroumpas                      
Phone: +30-1-772-1602, Fax: +30-1-772-1442
e-mail: {vkante,kpatro}@dblab.ece.ntua.gr

School of Electrical and Computer Engineering
Computer Science Division
National Technical University of Athens
Zographou, 157 73 Athens, Greece

[ Back ]