Loading...
Please wait, while we are loading the content...
Similar Documents
Optimised Short Block Codes for Iterative Joint Source and Channel Decoding in H . 264 Video Telephony
| Content Provider | Semantic Scholar |
|---|---|
| Author | Hanzo, Lajos Hanzo |
| Copyright Year | 2009 |
| Abstract | In this paper we propose a family of Short Block Codes (SBCs) designed for guaranteed convergence in soft-b i assisted iterative joint source and channel decoding, which f acilitate improved iterative soft-bit source decoding (SBSD) and cha nnel decoding. Data-Partitioned (DP) H.264 source coded video i s used to evaluate the performance of our system using SBC assisted SBSD in conjunction with Recursive Systematic Con volution Codes (RSC) for transmission over correlated narrowband Rayleigh fading channels. The effect of different SBC schem es having diverse minimum Hamming distances ( dH,min) and code rates on the attainable system performance is demonstrated , when using iterative SBSD and channel decoding, while keepi ng the overall bit-rate budget constant by appropriately partitioning the total available bit rate budget between the source and ch annel codecs to improve the overall BER performance and to enhance the objective video quality expressed in terms of Peak Signa lto-Noise Ratio (PSNR) . EXIT Charts were used for analysing the attainable system performance. Explicitly, our experi mental results show that the proposed error protection scheme usin g rate1 3 SBCs havingdH,min = 6 outperforms the identical-rate SBCs having dH,min = 3 by about 2.25 dB at the PSNR degradation point of 1 dB. Additionally, an Eb/N0 gain of 9 dB was achieved compared to the rate5 6 SBC having dH,min = 2 and an identical overall code-rate. Furthermore, anEb/N0 gain of 25 dB is attained at the PSNR degradation point of1 dB, while using iterative soft-bit source and channel decodingwith the aid of rate3 SBCs relative to the identical-rate benchmarker. I. M OTIVATION AND BACKGROUND Reliable transmission of multimedia source coded streams over diverse wireless communication networks, constitute s a challenging research topic [1, 2]. Since the early days of wi reless video communications [3–6] substantial further advan ces have been made both in the field of proprietary and standardbased solutions [7, 8]. Furthermore, the joint optimisatio n of different functions such as joint source and channel decodi ng (JSCD) gained considerable attention. The family of JSCD schemes often relies on exploiting the residual redundancy in the source-coded bit-stream. Fingscheidt and Vary [9, 10 ] proposed SBSD to exploit the natural residual redundancy of the source-coded bit-stream for improving the convergence of Iterative Source-Channel Decoding (ISCD) [11, 12]. Howeve r, only moderate residual redundancy is left in the source code d 1PSNR is the most widely used and simple form of objective vide o quality measure, which represents the ratio of the peak to peak signa l to the root mean squared noise [1]. Copyright (c) 2008 IEEE. Personal use of this material is per mitted. However, permission to use this material for any other purpo ses must be obtained from the IEEE by sending a request to pubs-permissi on @ieee.org. bit-stream, when using advanced state-of-the-art coding t echniques. Therefore we propose to deliberately impose additi onal redundancy on the source coded bit-stream with the aid of the novel class of SBCs proposed. In our experimental setup the H.264/AVC video codec [13] is used to encode the input video sequence and to generate the source coded bitstream. The H.264/AVC codec employs heterogeneous Variable Length Coding (VLC) and predictive coding techniques to achieve a high compression efficiency, which makes the compressed bit stream susceptible to transmission errors [1]. A single bit rror in the coded stream may corrupt the decoding of numerous future codewords. Moreover, due to predictive coding the effects of channel errors may affect the neighboring video blocks due to error propagation. Therefore the transmissio n of compressed video over wireless systems is a challenging tas k. Various error resilient schemes have been proposed in [1], in order to alleviate these problems, but the price paid is a potential reduction of the achievable compression efficien cy and increase in computational complexity. An iterative joi nt source-channel decoding procedure inspired by the concept of serial concatenated codes was presented in [14]. A symbolbased soft-input a posteriori probability (APP) decoder wa s presented in [15], where the residual redundancy was exploi ted for improved error protection. Instead of the traditional s erial concatenation of the classic Variable Length Codes (VLC) with a channel code, a parallel concatenated coding scheme was presented in [16], where the VLCs were combined with a turbo code. On the other hand, a novel irregular variable len gth coding (IrVLC) scheme designed for near-capacity joint sou rce and channel coding was proposed in [17]. Likewise, [18] advocated the employment of state-of-the-art High-Speed Pac ket Access (HSPA)-style [19] burst-by-burst adaptive transce ivers for interactive cellular and cordless video telephony, whi ch are capable of accommodating the time-variant channel qual ity fluctuation of wireless channels. An iterative source an d channel decoding aided Irregular Convolutional Coded (IRC C) videophone scheme using Reversible Variable-Length Codes (RVLC) and the Maximum A-Posteriori (MAP) [20] detection algorithm was proposed in [21]. The performance analysis of soft bit assisted iterative joint source and channel decodi ng was presented in [22], where Differential Space-Time Spreadin g (DSTS) aided Sphere Packing (SP) modulation was invoked which dispensed with channel estimation and provided both spatio-temporal diversity as well as a multi-user support. Furthermore, a joint source-channel decoding method based Authorized licensed use limited to: UNIVERSITY OF SOUTHAMPTON. Downloaded on May 6, 2009 at 07:37 from IEEE Xplore. Restrictions apply. This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. on the MAP algorithm was proposed by Wang and Yu [23]. Instead of the well known convolutional coded ISCD, an ISCD based on two serial concatenated short block codes was proposed by Clevorn et al. [24]. In the above-mentioned scheme a(6, 3) outer block code served as a redundant index assignment, while arate-1 block code was used as inner code. Similarly, Thobaben [25] provided the performance analysi s of a rateR = 4 5 single parity check code used for protecting the quantised source symbols relative to specifically desig n d VLCs. Joint source channel coding schemes employing a rate R = 4 5 linear block code for mapping the quantised source symbols to a binary representation were combined with an inner irregular channel encoder in [26]. An optimised bit ra te allocation scheme using a rate r = 12 inner channel encoder along with k = 3 to k = 6 source mapping was proposed in [27], and its performance was evaluated relative to conventional ISCD using a rater = 12 recursive non-systematic convolutional (RNSC) inner code. The Turbo DeCodulation scheme presented in [28, 29] consisted of two iterative loop s. The inner loop was constituted by the two components of Bit-Interleaved Coded Modulation using Iterative Decodin g (BICM-ID) and the outer loop by the Iterative Source-Channe l Decoding (ISCD) scheme. By contrast, in [30] a short block code based redundant index assignment and multi-dimension al mapping were used to artificially introduce redundancy and a single iterative loop was employed. Similarly, Clevorn et al. [31] presented a new design and optimisation guidelines for the ISCD’s performance improvement using the concept of redundant index assignment with specific generator matrice s. In contrast to this background, where specific mapping examples were provided for iterative source and channel codin g, we present powerful yet low-complexity algorithms for SBCs , which can be used to generate SBCs for a variety of mapping rates associated with diverse dH,min values that are applicable to wide-ranging multimedia services. Additionally, inste ad of modelling the sources with the aid of their correlation, the practically achievable interactive video performance tre nds are quantified when using state-of-the-art video coding techni ques, such as H.264/AVC. More explicitly, instead of assuming a specific source-correlation model, we based our system desi gn examples on the simulation of the actual H.264/AVC source coded bit-stream. The SBC-aided SBSD scheme was utilised for protecting the H.264 coded bit-stream using Recursive Systematic Convolution (RSC) codes [20]. The SBC coding scheme was incorporated by carefully partitioning the tota l available bit-rate budget between the source and channel codecs, which results in an improved performance when ISCD is employed. Explicitly, anEb/N0 gain of 25 dB is attained using ISCD, when employing a rate1 3 SBC scheme in comparison to a realistic identical-rate benchmarker disp en ing with the SBC scheme. The rest of the paper is organised as follows. In Section II we portray our system model along with an overview of the H.264 encoded bit-stream and SBSD. An overview of 2The superscript ⋆ was adopted from [27] ISCD is provided in Section III along with our proposed SBC coding algorithms, accompanied by a design example. Section IV provides the EXIT chart analysis of our proposed system model, while the performance of the proposed system is characterised with the aid of our simulation results in Section V. Finally we offer our conclusions in Section VI. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://eprints.soton.ac.uk/267333/1/H264_Video_Technology.pdf |
| Alternate Webpage(s) | http://eprints.soton.ac.uk/267333/1/H264_Video_Technology.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
