Energy efficient cooperative node management for wireless multimedia sensor networks

In Wireless Multimedia Sensor Networks (WMSNs) the lifetime of battery operated visual nodes is limited by their energy consumption, which is proportional to the energy required for sensing, processing, and transmitting the data. The energy consumed in multimedia sensor nodes is much more than in the scalar sensors; a multimedia sensor captures images or acoustic signals containing a huge amount of data while in the scalar sensors a scalar value is measured (e.g., temperature). On the other hand, given the large amount of data generated by the visual nodes, both processing and transmitting image data are quite costly in terms of energy in comparison with other types of sensor networks. Accordingly, energy efficiency and prolongation of the network lifetime has become a key challenge in design and implementation of WMSNs. Clustering in sensor networks provides energy conservation, network scalability, topology stability, reducing overhead and also allows data aggregation and cooperation in data sensing and processing. Wireless Multimedia Sensor Networks (WMSNs) are characterized for directional sensing, the Field of View (FoV), in contrast to scalar sensors in which the sensing area usually is uniform and non-directional. Therefore, clustering and the other coverage-based techniques designed for WSNs, do not satisfy WMSNs. In WMSNs, sensor management policies are needed to assure balance between the opposite requirements imposed by the wireless networking and vision processing tasks. While reducing energy consumption by limiting data transmissions is the primary challenge of energy-constrained visual sensor networks, the quality of the image data and application, QoS, improve as the network provides more data. In such an environment, the optimization methods for sensor management developed for wireless sensor networks are hard to apply to multimedia sensor networks. Such sensor management policies usually employ the clustering methods which form clusters based on sensor neighbourhood or radiocoverage. But, as it was mentioned, because of the main difference between directional sensing region of multimedia sensors and the sensing range of scalar sensors, these schemes designed for WSNs, do not have efficiency for WMSNs. Moreover, sensor management strategies of WSNs do not consider the eventdriven nature of multimedia sensor networks, nor do they consider the unpredictability of data traffic caused by a monitoring procedure. This thesis, first, present a novel clustering mechanism based on the overlapping of the FoV of multimedia nodes. The proposed clustering method establishes clusters with grouping nodes that their FoVs overlap at least in a minimum threshold area. Two styles of cluster membership are offered by the mechanism depending on the desired network application; Single Cluster Membership (SCM) and Multi Cluster Membership (MCM). The name of MCM comes from the fact that a node may belong to multiple clusters, if its FoV intersects more than one cluster-head (CH) and satisfies the threshold area while in SCM each node belongs to exactly one cluster. Then, the proposed node management schemes designed for WMSNs are presented; the node selection and scheduling schemes manage the acts of the multimedia sensor nodes in a collaborative manner in clusters with employing the mentioned clustering method. Intra-Cluster Cooperation (ICC) and Intra&Inter-Cluster Cooperation (IICC) use the SCM and MCM clusters respectively. The monitoring period is optimized and the sensing region is divided among clusters and multimedia tasks are performed applying cooperation within and between clusters. The objective is conserving the residual energy of nodes to prolong the network lifetime. Finally, a hybrid architecture for WMSNs in order to energy efficient collaborative surveillance is proposed. The proposed mechanism employs a mixed random deployment of acoustic and visual sensor nodes. Acoustic sensors detect and localize the occurred event/object(s) in a duty-cycled manner by sampling the received signals and then trigger the visual sensor nodes covering the objects to monitor them. Hence, visual sensors are warily scheduled to be awakened just for monitoring the object(s) detected in their domain, otherwise they save their energy. Section B. 4 of Chapter I introduces the contributions of this thesis.