Algorithms and architectures for secure embedded multimedia systems
Pande, Amit (2010) Algorithms and architectures for secure embedded multimedia systems. PhD thesis, Iowa State University.
Full text available as:
Embedded multimedia systems provide real-time video support for applications in entertainment (mobile phones, Internet video websites), defense (video-surveillance and tracking) and public-domain (tele-medicine, remote and distant learning, trafﬁc monitoring and management). With the widespread deployment of such real-time embedded systems, there has been an increasing concern over the security and authentication of concerned multimedia data. While several (software) algorithms and hardware architectures have been proposed in the research literature to support multimedia security, these fail to address embedded applications whose perfor- mance speciﬁcations have tighter constraints on computational power and available hardware resources. The goals of this dissertation research are two fold: 1. To develop novel algorithms for joint video compression and encryption. The proposed algorithms reduce the computational requirements of multimedia encryption algorithms. We propose an approach that uses the compression parameters instead of compressed bitstream for video encryption. 2. Hardware acceleration of proposed algorithms over reconﬁgurable computing platforms such as FPGAs and over VLSI circuits. We use signal processing knowledge to make the algorithms suitable for hardware optimizations and reduce the critical path of circuits using hardware-speciﬁc optimizations. The proposed algorithms ensure a considerable level of security for low-power embedded systems such as portable video players and surveillance cameras. These schemes have zero or little compression losses and preserve the desired properties of compressed data in encrypted bitstream to ensure secure and scalable transmission of videos over heterogeneous networks. The proposed algorithms also support indexing, search and retrieval in secure multimedia digital libraries. This property is crucial not only for police and armed forces to retrieve information about a suspect from a large video database of surveillance feeds, but extremely helpful for data centers (such as those used by youtube, aol and metacafe) in reducing the computation cost in search and retrieval of desired videos.
Archive Staff Only: edit this record