On the Unequal Error Protection in Compressive Video Sensing For Low Power and Lossy IoT Networks

Abstract

Encoding and transmitting multimedia data over low-cost wireless sensors require a trade-off between energy and quality. Compressive Sensing (CS) has been proposed to transfer encoder complexity to the decoder and save energy at the sender, so it has been beneficial for recently developed Internet of Things (IoT) technologies. On the other hand, in practical applications of IoT, such as Low Power and Lossy Networks (LLNs), data is prone to error, which affects the quality at the receiver side. In order to solve this issue, error control mechanisms have been employed, while they are energy-consuming. Therefore, finding an efficient method for error protection in CS encoded videos is crucial in IoT lossy networks. This paper investigates the impact of Unequal Error Protection (UEP) on different frame types for a CS video setting in IoT lossy networks. Experimental results show that making a less complex and energy-efficient encoder can be done by not treating all frames with the same error control mechanism. Instead, having no or less complicated error control methods for non-key (P and B) frames but protecting I-frames is a better solution. Compared to equal error protection for all frame types, our method results in around 20% better SSIM (Structural Similarity Index) for CS video delivery over IoT LLNs.