Real-time video encryption scheme based on multi-round confusion-diffusion architecture

With the rapid development of network and multimedia technologies, the widespread use of images in social networks, military applications, and other fields has led to substantial demands for image encryption. To protect images from potential threats, most protocols perform multiple rounds of confusion and diffusion until a satisfactory security level is attained. However, this strategy is time-consuming and unable to meet the requirements for real-time video encryption. Consequently, many works realize video encryption by simplifying the encryption process, e.g. employing a single round of confusion and diffusion to encrypt each frame. To enhance security to the level of image encryption while ensuring real-time performance, this paper proposes a real-time video encryption protocol based on multi-round confusion-diffusion architecture. It splits the frame into subframes, uses a set of threads to concurrently perform confusion and diffusion. The statistical and security analyses demonstrate the proposed protocol exhibits exceptional statistical properties and is resistant to various attacks. The encryption speed evaluation shows that our method significantly enhances computational efficiency, achieving latency-free encryption of 512×512 24FPS videos using the Intel Xeon Gold 6226R@2.9GHz CPU. Even with the execution of four rounds of confusion and six rounds of diffusion operations on each frame, the average encryption time remains below 30ms.