National and Kapodistrian University of Athens
Distributed signal processing and data fusion methods for large scale wireless sensor network applications
Abstract
dc:descriptionIn this Dissertation we study the problem of tracking the boundary of a continuous object (e.g. a hazardous diffusive phenomenon) and predicting its local and global spatio-temporal evolution characteristics using large-scale Wireless Sensor Networks (WSNs). We introduce a practical WSN-based approach consisting of two main components: a) An asynchronous collaborative in-network processing algorithm that estimates, using dynamically formed node triplets (clusters), local front model evolution parameters (orientation, direction and speed) of the expanding continuous object, and b) an algorithm that reconstruct the overall hazard's boundary by combining the produced local front estimates as they are becoming available to a fusion center. Based on the estimated uncertainties of local front model parameters, the reconstruction can provide for each point of the considered area the probability to be reached by the hazard’s front. Extensive computer simulations demonstrate that the proposed algorithm can estimate accurately the evolution characteristics of complex diffusive continuous objects, while it remains robust to sensor node and communication link failures. Finally, we show that it can track with accuracy the evolution of continuous objects with complex shapes, using a relatively small number of potentially distorted local front estimates.
Degree
thesis:*- Grantor dc:publisher
- National and Kapodistrian University of Athens
- Year dc:date
- 2014
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Manatakis, Dimitrios
- Μανατάκης, Δημήτριος
Subjects
dc:subject × 16- Κατανεμημένη επεξεργασία σήματος
- Σύντηξη δεδομένων
- Ασύρματα δίκτυα αισθητήρων
- Εκτίμηση παραμέτρων
- Μηχανική μάθηση
- Παρακολούθηση συνεχών αντικειμένων
- Distributed signal processing
- Data fusion
- Wireless sensor networks
- Parameters estimation
- Machine learning
- Continuous object tracking
- Φυσικές Επιστήμες
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Natural Sciences
- Computer and Information Sciences
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/42294