Adaptive Video Transmission over MIMO Channels

Abstract

Communication over wireless channels has matured considerably in the last decade. While voice and low end data content can be carried reliably and efficiently over contemporary wireless chan- nels, communication of video content over these channels still remains an open research problem. Video data is complicated for transmission over wireless media because of its high band- width and low delay requirements. Furthermore, high compu- tational cost of video encoding and decoding is prohibitive for battery- and resource-constrained mobile devices. Due to their high spectral efficiency (high data rates) and good diversity char- acteristics (low BERs), wireless systems with Multiple-Input- Multiple-Output (MIMO) physical layers are quite well-suited to mitigate the above problems. The two complementary ap- proaches employed for transmission of data over MIMO systems are: Spatial Multiplexing (SM) and Space Time block Coding (STBC). SM maximizes throughput whereas STBC provides di- versity gains. The benefits of STBC and SM can also be com- bined by adaptively switching between the two schemes in SNR regions where one scheme outperforms the other. In this the- sis, we study adaptive video transmission over MIMO channels. We carry out this research in two steps. Firstly, we investi- gate a receiver-centric comparison of different video sequences over STBC and SM systems. Secondly, we propose cross-layer adaptive scheme that can intelligently switch between SM and STBC and can significantly improve the quality of video trans- i ii mission overMIMO channels. In the receiver-centric comparison we use empirical SNR versus PSNR performance evaluations of H.264 video sequences to show that reasonable video qual- ity can be provided using the basic (SM and STBC) diversity- multiplexing schemes. We identify three distinct channel SNR regions in which SM and STBC PSNR performance changes based on varying motion and channel characteristics. We ob- serve that the boundaries of the classified regions are flexible and depend largely on the motion intensity of the video con- tent, which directly impacts decoding performance. Using our deductions from the receiver-centric evaluation, out proposed cross-layer adaptive video transmission scheme uses the motion intensity of the video content and the received channel SNR to adaptively switch between SM and STBC. Since low BERs does not necessarily translates into improved video quality, our goal is to devise a switching scheme that takes into consideration characteristics of video data in order to achieve better PSNR. We show that a measure of motion intensity on the applica- tion layer of the transmitter should be used to compute the switching SNR–the SNR after which SM outperforms STBC. Experimental results show that the proposed scheme provides significantly better PSNR than standalone SM and STBC and existing content-unaware switching mechanisms.