Adaptive encoding scheme for real-time video streaming over mobile networks

Advances in mobile technologies have continuously improved the capability and data access of mobile devices. High resolution camera, large multi-touch screen and improving data rate make video communication more ready to replace voice communication on mobile phones. However, connectivity and data rate are not guaranteed in mobile networks which make it hard to maintain a jitter-free streaming of live video on mobile devices. In this paper, we propose an adaptive encoding scheme to maintain a jitter-free video streaming with minimum quality degradation under wireless network. For each video frame, we define a region of interest (ROI) where the receiver has low tolerance on quality degradation and to try to maintain the region quality. Experimental results demonstrated the effectiveness of the proposed adaptive encoding scheme.

I. Introduction

Advances in mobile technologies have opened up a new horizon for mobile computing applications. Improving processing power, camera and screen resolution, network bandwidth, location and orientation sensors make mobile phones go far beyond simple voice communicating devices. Video communication, like videoconferencing, is more ready to replace voice communication in the future. Video communication typically involves live-video streaming, i.e. video signals are captured by a built-in camera, encoded into a compressed format, transmitted over a network and then finally decoded and rendered on the receiver screen. Besides videoconferencing, live-video streaming could also be used in veracious remote assistance services. A paramedic, after on arriving the scene, can use a mobile phone to capture and transmit the situation of the scene / patients to doctors located in remote hospitals. This will enable the doctors to give instruction that the on-site paramedics could not provide. However, the primary requirement of the streaming is to maintain a jitter-free video playback on the receiver side and it requires a stable bandwidth provided by the network to transport video frames in a timely manner. However, typical mobile networks are error-prone and present a major challenge if a jitter-free video stream is to be maintained. A lot of researchers have proposed various efficient technologies and approaches to meet this requirement [6], [7], [9], [10], [11]. In addition, the bandwidth provided is limited and fluctuates widely, with frequent disconnection. Furthermore, ever increasing camera resolution and screen size places additional demand on the bandwidth requirement for video streaming. In this paper, we propose an adaptive encoding scheme for live video streaming over a mobile network. In videoconferencing, certain area may draw more attention than the other areas, i.e. the speaker face in videoconferencing and the wound of the patient in the ambulance services. Since a user generally has more tolerance on quality degradation outside the region of interests, when effective bandwidth provided by the network is reduced, bandwidth allocated for transmitting those areas can be reduced.