Adapting quantization offset in multiple description coding for error resilient video transmission q Viswesh Parameswaran, Avin Kannur, Baoxin Li * Department of Computer Science and Engineering, Arizona State University, Tempe, AZ 85281, USA article info Article history: Received 21 August 2008 Accepted 4 July 2009 Available online 12 July 2009 Keywords: Multiple description coding (MDC) Video streaming H.264/AVC encoder Error resilient video transmission Wireless ad hoc networks Video quality Video over wireless Quantization offset abstract Multiple description coding (MDC) provides an excellent error resilient approach for transmitting video over wireless ad hoc networks. In this paper, we propose an improvement to this scheme by jointly selecting the quantization offsets in different paths to achieve the best overall video quality at the deco- der. The statistical distribution of the transform coefficients in the encoded video sequence is parameter- ized using a Laplacian model. The optimal offset values are computed by solving a multi-level optimization problem based on the statistics of the transform coefficients and the individual path failure probabilities. In order to reduce the computational complexity, the encoding modes for the motion vec- tors and transform coefficients are collected in the first step. In the second step, the model parameter is calculated from the transform coefficients and the offset search is performed when the model parameter between frames deviates beyond the pre-set threshold. In the final step, the stored modes and the respec- tive offset values are directly used for encoding the two bit streams. The simulation results using H.264/ AVC system confirm the advantages of the proposed approach under different packet loss conditions. Ó 2009 Elsevier Inc. All rights reserved. 1. Introduction The past decade has witnessed rapid growth in the develop- ment and deployment of computationally intensive multimedia applications on wireless devices. This can be attributed to the fol- lowing reasons: – (1) the improvements in semiconductor technol- ogies resulting in the development of 45 nm integrated chip fabrication process, (2) the developments in wireless technologies enabling information transfer at speeds ranging 2 Mb/s in the case of 3G networks, and (3) the evolution of video compression stan- dards enabling high quality video transmissions at low bit rates. Some of the cell phones available in market are already equipped with digital video streaming and conferencing functionalities along with the fundamental voice communication feature. The antici- pated growth in mobile social networking will further increase the utility of multimedia transfer over cell phones. The typical vi- deo applications running on these devices can be categorized on the basis of their underlying network, delay requirements and the desired video quality. The network used for video transfer could be fixed wire line, fixed wireless, ad hoc wireless or a combi- nation of wired and wireless networks. The delay parameters that need to be considered for system design are the average end-to- end delay (known as latency) and the delay variation (known as jit- ter). The video quality can be specified in terms of the required spa- tial and temporal resolutions. For instance, consider the case of one-way video streaming over internet. In this application, the video is compressed offline and is stored in a streaming server. The user connects to the server using a suitable protocol such as real-time streaming protocol (RTSP). Since the video is encoded offline, there is no delay constraint at the encoder. The decoder stores the received frames in a buffer and starts playing them after an initial play-out delay. In this case the latency can be fairly large, but jitter should be limited so that the video can be played out smoothly. The maximum jitter that can be tolerated is determined by the size of the playout buffer. The desired video quality in the case of this application is deter- mined by the end user terminal and the speed of the linking network. Another application scenario that can be envisioned is interac- tive video conferencing over cell phones. In this case we need to capture and deliver video in real time. In order for the two-way communication to be effective, we need to impose stringent delay constraints both at the encoder and the decoder. The system has to be designed in such a way so as to minimize latency and jitter. The buffers at the encoder and decoder should be of limited size to minimize latency. The underlying network in this case is wireless and hence the application is constrained by limited band-width 1047-3203/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.jvcir.2009.07.003 q This research was supported in part by ARO Grant W911NF-06-1-0354. The information reported here does not reflect the position or the policy of the federal government. * Corresponding author. E-mail addresses: vparames@asu.edu (V. Parameswaran), akannur@asu.edu (A. Kannur), baoxin.li@asu.edu (B. Li). J. Vis. Commun. Image R. 20 (2009) 491–503 Contents lists available at ScienceDirect J. Vis. Commun. Image R. journal homepage: www.elsevier.com/locate/jvci