The use and further development of the multimedia-based services and apps fueled an exponential rise in social multimedia traffic. In that context, the realization of all fundamental needs of the social multimedia networks, which include the scalability, dependability, quality of service and quality of information (QoI), depends critically on safe transfer of the data (QoS). As a result, a paradigm based on trust is widely required for multimedia analytics to satisfy growing user demands and provide more immediate and useful insights. In this context, software-defined networks are essential; nevertheless, their ability to support effective network control and regulation is constrained by several variables, including the runtime security as well as the energy-aware networking. Thus, a hybrid DL-based anomaly detection approach for the suspicious flow detection in contexts of the social multimedia has been developed for the purpose of improving SDN's dependability. It consists of two modules: a module for anomaly detection that uses a hybrid convolutional and LSTM network to identify aberrant activity and a module for end-to-end data transport that complies with the stringent QoS requirements of the SDN, such as the low latency and high bandwidth. In order to demonstrate the suggested scheme's efficiency and effectiveness in terms of anomaly detection and data transmission that is necessary for social multimedia, it was empirically evaluated on benchmark and real-time datasets.
