IMPACT-2013 Performance Investigation of Application Layer Unequal Error Protection for Embedded Video Bitstream Mohd Ayyub Khan, Athar Ali Moinuddin, Ekram Khan Department of Electronics Engineering, Aligarh Muslim University, Aligarh, UP, INDIA ayyub.alig@gmail.com atharamoin@gmail.com ekhan67@gmail.com Abstract— The wavelet-based video coders are excellent in providing the fine granular scalability (FGS), i.e. progressive bitstream, which can be used to multicast the video transmission to different users requiring different bit-rate, resolution and frame rate, from single coded bitstream. However, transmission of these bitstreams over erroneous channel is a challenging task. In this paper, application layer unequal error protection (UEP) using RS codes to embedded video bitstream transmitted over Additive White Gaussian Noise (AWGN) channel is investigated. The UEP of embedded bitstream exploits the non-uniform importance of the bits, in the reconstruction of the video. The bitstream is partitioned into two substreams namely, high priority (HP) and low priority (LP), depending upon their importance and sensitivity to channel errors. Then FEC based error protection is provided to them according to their priority. The results show that, under poor channel conditions, the UEP scheme improves the quality of the reconstructed video over EEP scheme for AWGN channel. I. INTRODUCTION In recent years, the demand for video based multimedia services using portable devices like cell-phones, smart-phones, tablets, etc., are increasing at a very fast rate. This requires transmission of compressed video (H.264/AVC, MPEG-4 coded video) over wireless and IP networks, which is a challenging task. This is due to the fact that these channels are erroneous in nature either because of fading and interference or high congestion at the routers. Usually, fading results in loss of packets at the receiver while router drops the packets to control the congestion. In addition, the compressed video is highly correlated, and even single bit error may result in catastrophic failure in the reconstruction of received video. Therefore, an error resilient technique is required to combat the effect of errors to the coded bitstream. Over the years, several error resilient techniques have been used to counter the effect of channel distortions [1]. These include Forward Error Correction (FEC), retransmission using Automatic Repeat Request (ARQ), error resilient coding, hierarchical modulation, multiple descriptive coding, error concealment, etc. Another approach of providing robustness against noisy channel on the basis of non-uniform importance of data in [2], called unequal error protection (UEP), was firstly proposed. UEP was developed to protect video data having non-uniform importance generated by layered video coder. For example, simple layered video coder would generate two layers namely: base layer (BL) and enhancement layer (EL). The base layer is more important than enhancement layer. For the reconstruction of video, decoding the base layer only will provide a coarse quality while adding the enhancement layer will further improve the reconstruction quality of the video. Therefore, base layer may be protected more strongly than enhancement layer resulting in UEP of the coded video. In [3], UEP for convolutional codes has been suggested. In [4], message-wise UEP scheme using LDPC codes was proposed, which provides the large protection to high- significance bits while small protection to low-significance bits. However, above analysis were carried out over statistically independent data, however, video coded data are highly correlated and have characteristics different than ordinary data. In [5], UEP is provided to embedded bitstream generated by the derivative of the Set Partitioning in Hierarchical Trees (SPIHT). In this scheme, the wavelet coded image bitstream is divided into three substreams, depending upon their sensitivity to channel errors. The scheme is applied to image only. Therefore, it is worthwhile to investigate the performance of UEP over wavelet based video bitstream, which is the main motivation of this work. In this paper, UEP is used for protection of Wavelet- Block-Tree-Coder (WBTC) coded video bitstream over AWGN channel to overcome the effect of channel errors. WBTC is a wavelet based coder [6][7], which combines the feature of SPIHT [8] and SPECK [9] in a single algorithm. The motivation for using WBTC is that after wavelet 978-1-4799-1205-6/13/$31.00 ©2013 IEEE 31