Real-time Multi-resolution Data Transmission over Correlated Fading Channels using Hierarchical Constellations

We study in this paper an integrated packet scheduling and unequal error protection strategy for the transmission of packetized multi-resolution (layered) image/video data over time varying wireless channels. Different transmission priorities are given to different layers of video/image data not only by using hierarchical constellation (known also as nonuniform, asymmetric, multi-resolution constellation) at the physical layer but also by a buffer adaptive packet scheduling protocol at the medium access control (MAC) layer. Basically, according to the current buffer occupancy and channel state, the scheme dynamically selects packets from different resolution levels to transmit for the current transmission. The bits from the selected packets are assigned to different hierarchies of a hierarchical 4/16-quadrature amplitude modulation (QAM). In our proposed transmission framework, the transmit power level and the constellation priority parameter are also selected dynamically according to the quality of the wireless channel and scheduling state. In order to minimize the distortion in the received image/video with a given power budget, a joint optimization of the packet scheduling protocol, the transmit power level, and the constellation priority parameter is proposed. The power-distortion trade-offs curve is plotted for a given fading rate. We also compare the proposed scheme with a scheme employing uniform signal constellation. The comparison shows that in order to achieve a given quality of the received image/video, the scheme employing hierarchical QAM constellation always requires less power than the scheme employing uniform constellation.