A Cross-Layer Approach for Maximizing Visual Entropy Using Closed-Loop Downlink MIMO

Hindawi Publishing Corporation EURASIP Journal on Advances in Signal Processing Volume 2008, Article ID 864606, 14 pages doi:10.1155/2008/864606 Research Article A Cross-Layer Approach for Maximizing Visual Entropy Using Closed-Loop Downlink MIMO Hyungkeuk Lee, Sungho Jeon, and Sanghoon Lee Wireless Network Laboratory, Yonsei University, Seoul 120-749, South Korea Correspondence should be addressed to Sanghoon Lee, Received 1 October 2007; Revised 27 March 2008; Accepted 8 May 2008 Recommended by David Bull We propose an adaptive video transmission scheme to achieve unequal error protection in a closed loop multiple input multiple output (MIMO) system for wavelet-based video coding. In this scheme, visual entropy is employed as a video quality metric in agreement with the human visual system (HVS), and the associated visual weight is used to obtain a set of optimal powers in the MIMO system for maximizing the visual quality of the reconstructed video. For ease of cross-layer optimization, the video sequence is divided into several streams, and the visual importance of each stream is quantified using the visual weight. Moreover, an adaptive load balance control, named equal termination scheduling (ETS), is proposed to improve the throughput of visually important data with higher priority. An optimal solution for power allocation is derived as a closed form using a Lagrangian relaxation method. In the simulation results, a highly improved visual quality is demonstrated in the reconstructed video via the cross-layer approach by means of visual entropy. Copyright © 2008 Hyungkeuk Lee et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 1. INTRODUCTION The ongoing broadband wireless networks have attractive advantages for providing a variety of multimedia streaming applications while guaranteeing the quality of service (QoS) for mobile users. Nevertheless, many limitations for adapting the magnificent growth of multimedia traffic into expensive and capacity-limited wireless channels continue to exist. The multiple input multiple output (MIMO) system is capable of increasing channel throughput drastically by using multiple transmit and multiple receive antennas [1, 2]. Since the MIMO channel is composed of multiple parallel subchannels with different quality, more efficient radio resource management can be developed by exploiting such different channel characteristics. If higher and lower quality subchannels are used for more and less important data, respectively, from the perspective of cross-layer optimization, a better performance could be expected. Some recent papers have highlighted issues of cross-layer optimization for achieving a better quality of source over a capacity-limited wireless channel [3–7]. If source-dependent information exchanges across the top and bottom protocol layers are used, more improved performance can be obtained even if the exchanges may not be available in traditional layered architectures in [3]. The authors in [4] presented a high-level framework for resource-distortion optimization, that jointly considered factors across the network layer, including source coding, channel resource allocation, and error concealment. In [5], a framework of cross-layer design for supporting delay critical traffic over ad-hoc wireless networks was proposed and its benefits for video streaming were analyzed. In [7], a modified moving picture experts group (MPEG)-4 coding scheme was employed for progressive data transmission by controlling the number of subcarriers over a multicarrier system. Besides, the authors in [8–15] exploited joint transmission and coding schemes over MIMO systems using not only the layered coding, but also the multiple description coding (MDC). In [8], an unequal power allocation scheme for transmission of joint photographic experts group (JPEG) compressed images employing spatial multiplexing was proposed, so a significant image quality improvement was achieved compared to other schemes. Similarly, in [9], the unequal spatial diversity scheme was proposed for providing unequal error protection, which was based on 2 EURASIP Journal on Advances in Signal Processing (a) PSNR = 22.3 Visual entropy = 8538.0 (b) PSNR = 23.6 Visual entropy = 10490.0 (c) PSNR = 25.1 Visual entropy = 11812.5 (d) PSNR = 22.2 Visual entropy = 4911.2 (e) PSNR = 23.6 Visual entropy = 5232.2 (f) PSNR = 25.7 Visual entropy = 6386.6 Figure 1: Quality assessment using PSNR versus visual entropy. the combined use of turbo codes and space-time codes. It could also provide a reduction in average transmission time and a image quality improvement compared with no spatial diversity, but the criteria was not suggested. Authors in [10] presented the gains arising from transmitting MDC over spatial multiplexing (SM) systems. Authors in [11] showed that the layered coding might outperform MDC under certain conditions when an error-free environment or an environment with a very low-error rate can be guaranteed for the base layer. Nevertheless, it is presented that MDC can be one of the realistic MIMO transmission scenarios as good as the layered coding can in [12]. Authors in [13] observed that the general water-filling power allocation, while optimizing the capacity of MIMO singular value decomposition (SVD) system, may not be optimal for video. From the perspective of cross-layer optimization, the major drawback in the previous research is the lack of the specific criteria defining the importance of each information bit. Moreover, the heuristic algorithm without the use of a mathematical proof is only presented. In order to adapt a bulky multimedia traffic to a capacity-limited wireless channel, it is necessary to generate layered video bitstreams and then to transmit more visually important data to higher quality subchannels and vice versa. Even if it is easy to conceive such idea, the main issue is how the radio resource control can be conducted based on which criterion. The most widely used quality criterion peak signal-to-noise ratio (PSNR) does not characterize the quality of the visual data perfectly. Figure 1 illustrates the defect in the PSNR value. Even though, the PSNR values shown in Figures 1(a), 1(b), and 1(c) are approximately the same as those shown in Figures 1(d), 1(e), and 1(f), respectively, the visual qualities for them are significantly different because the PSNR criterion cannot determine where distortion comes from. Therefore, the PSNR as a quality assessment does not accurately represent visual quality. However, the PSNR is known as the dominant quality assessment because, in spite of this defect, no clear quality criterion exists as an alternative. Therefore, the current technical limitation lies in the lack of quality criteria for evaluating the performance gain attained by the cr (...truncated)