Thesis : Data reduction techniques for Wireless Sensor Networks using mathematical models

Recently, an enormous revolution has occurred in systems and wireless communications while wireless sensor networks (WSNs) have become an interested research topic due to their main role in the cooperation of humans, environment, and machines. WSNs are very powerful tools that are employed in several fields, such as: military, health, smart homes and cities, agriculture, natural disaster prevention, etc. In general WSN is composed of several sensors that communicate together in order to transmit the sensed measures to the base stations. The interconnected sensor nodes are equipped with limited resources in terms of battery, memory, CPU, bandwidth and power consumption.

The main factor affecting nodes’ lifetime is their limited battery energy. This limitation in terms of resources, in addition to the fact that those sensors are randomly deployed may lead to a node failure or attack. As sensor nodes are generally battery-powered devices, the critical aspects to face concern how to enhance the life of battery in the sensor nodes, so that the network lifetime can be extended to reasonable times. Therefore, energy conservation is the principal issue in the implementation of systems based on WSNs. The major challenges faced by the network are the limited resources of the nodes that do not handle the huge amount of data collection and transmission, in addition to the security issues of the node’s information that are transferred in untrusted environment.

Therefore, researchers in WSNs focus on minimizing the data communication in order to prolong the lifetime of the network, while protecting the quality of the data with low cost security strategies.

In this thesis, we focus on data reduction and security in wireless sensor networks. In the data reduction phase three phases are proposed for energy efficiency improvement based on aggregation, correlation and prediction of the collected data. While in the last part we introduced a data aggregation security technique for WSNs that is suitable with their resource constraints. The network architecture presented in our techniques is a based tree topology. In this network, each sensor node sends its sensed measures to its aggregator, which in his turn sends this data to the base station. In our  work, we present energy efficient techniques for wireless sensor networks to decrease the data transmitted through the network with a security approach to ensure a healthy data reception.

Thesis jury composed of :

- M. Pierre Spitéri, professeur, INP Toulouse
- M. Hassan Artail, professeur, American University of Beirut
- M. Haidar Safa, professeur, American University of Beirut
- M. Mahmoud Barhamgi, Maître de conférences, Université Lyon 1
- M. Raphaël Couturier, Professeur, Université Bourgogne Franche-comté
- M. Abdallah Makhoul, Maître de conférences, Université Bourgogne Franche-comté
- M. Abbas Hijazi, professeur, Université Libanaise
- Mme. Samar Tawbi, Maître de conférences, Université Libanaise

Keywords : Wireless Sensor Networks, Multivariate data reduction, Similarity and Distance Functions, Fitting functions, Polynomial function, Security.