AMSEP: Automated Multi-level Security Management for Multimedia Event Processing

Efficient event processing is fundamental for IoT systems handling critical data such as security surveillance and e-health. There is a need for data real time processing and accurate classification of events following their security level. These tasks are challenging in IoT context systems which handle a big volume of multimedia data. They require a large processing time and event consumers have different points of interest and multi-level access rights. In this paper, we propose AMSEP for multi-level security management of multimedia events. AMSEP automatically analyses incoming multimedia events, calculates their security levels following multi-level security policies in order to send them securely to their destinations. The proposed system is based on Deep Learning processing and on a publish/subscribe broker extension (Kafka). We show that mixing the two building blocks allows to fulfill three main challenges in IoT: (1) Accurate and real time event processing, (2) Dynamic adaptation to modify security policies and (3) event scalability and reliable routing of events from sources to destinations. (C) 2018 The Authors. Published by Elsevier Ltd.