Intelligent architecture for multimedia transmission in wireless sensor networks

With the availability of low-cost small multimedia sensors, Wireless Multimedia Sensor Networks (WMSNs) have been proposed to harvest and communicate event features in the form of multimedia, and support multiple applications such as multimedia surveillance, target tracking, environmental monitoring, and traffic management systems. These applications pose substantive challenges regarding quality-driven, energy-efficient and secure multimedia processing and communication in WMSNs (i.e., heterogeneous multimedia reliability assurance, tight Quality of Service expectations and energy efficiency). Given the fundamental resource constraints, security protection and reliability issues of these sensors, secure multimedia transmissions must be judiciously designed to reduce data redundancy and computational complexity, enhance error resilience, and provide security and perceptual integrity. This dissertation is focused on these challenges and issues in cross-layer architecture design to support efficient and secure multimedia transmission in wireless sensor network environments.This research work has proposed an innovative architecture to comprehensively include source coding, unequal resource allocation and security protection in a crosslayer design. The proposed approach supports secure collaborative transmissions, adaptive watermarking image authentication and Modified Discrete Cosine Transform (MDCT) based audio streaming with selective encryption in WMSNs. The cross-layer optimization complexity can hinder a wide application of this idea and therefore an intelligent architecture is incorporated to analyze and identify significant cross-layer parameters using supervised learning approaches. This architecture also uses only a few key cross-layer parameters to improve overall WMSNs performance with reduced complexity. Our simulation and experimental results demonstrate that the proposed intelligent cross-layer architecture achieves considerable gains for multimedia sensor networks performance in terms of resource efficiency, Quality of Services (QoS) and security.