Intégration du Cloud Computing pour le stockage et l’agrégation de données dans les réseaux de capteurs sans fil .

Abstract

This thesis deals with the integration of Wireless Sensor Networks (WSNs) in Cloud Computing. The goal is to solve the intrinsic problems posed by these networks in terms of data storage, limited energy of the batteries of sensors and access to the network (congestion). In fact, it is about improving the performance of WSNs for longer network lifetime and a large number of applications. Since data transmissions are the basis of the problems mentioned, we first started by studying the impact of their reduction using dual prediction and data aggregation techniques. Thus, we have developed a new prediction algorithm, called EADPS (Extended Adaptative Dual Prediction Scheme). Next, we compared the data aggregation to each of ADPS (Adaptive Dual Prediction Scheme) and EADPS. It turns out that aggregation is a much better technique than the ADPS scheme for small networks and regardless of the accuracy of the predictions. However, it only becomes seriously competitive with the EADPS scheme for large sizes of WSNs. We performed different simulations on real data using prediction alone, aggregation alone and then their combination. It appears that with the combination of these two techniques we have maximized the rate of reduction of transmissions. Based on these encouraging results, we designed and produced simulation software for the integration of RCSF into Cloud Computing.