JSCC scheme

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

1.1 Related Work

Communication is all about trying to convey as much information as possible over a given channel with as few errors as possible. JSCC is a robust and effective mechanism used for data communication over error pone and bandwidth limited wireless channels. In [[i]] a brief overview of the most prominent techniques of JSCC are given. It covers index assignment (IA), unequal error protection (UEP), co-optimised vector quantizing and IA (i.e. channel optimised vector quantizer ), and direct modulation organising schemes . JSCC methods are developed for image transmissions are presented in [[ii]][[iii]][[iv]][[v]]. Unlike image transmission for video transmission more intelligent and effective JSCC schemes are necessary due to its demand for higher bandwidth. A JSCC scheme utilised for wireless video transmissions over Code Division Multiple Access (CDMA) networks is presented in [[vi]]. In this work, a compressed video bit stream is transmitted over multiple-channel of wireless CDMA networks. Each video source layer is protected by a product channel code structure, where row coding is a combination of a rate-compatible punctured convolution (RCPC) and cyclic redundancy check (CRC). Reed-Solomon (RS) is applied for column coding. This JSCC effectively protects transmitted video over multipath fading channels. Usage of the JSCC for video broadcasting over a WiMAX network in to enhance the performance of TV transmission over internet protocol (IP) is illustrated in [[vii]]. In [[viii]], the JSCC is introduced for scalable video coding (H.264/SVC) with low-density parity-check (LDPC) channel coding. The experimental results suggest that the LDPC provides high degree protection to any scalable setting of the SVC. In [[ix]] authors have proposed a, a hybrid error control technique consisting of two types of mechanisms: forward error correction (FEC) and retransmission. The objective of this study is to optimise the trade-off between overhead bits due to an application of the FEC and the system delay due to an application of the ARQ to retransmit lost packets. JSCC scheme to minimise average distortion if channel conditions are known at a receiver side is proposed in [[x]]. In this study JSCC methods are designed for three different scenarios. In the first scenario JSCC scheme evaluates an optimal channel coding rate assuming the source coding rate is know. The second scenario finds an optimal source coding rate assuming the channel coding rate is know. The last scenario iteratively searches for an optimal bit rate between source and channel coding. Here a channel-optimised vector quantisation (COVQ) is applied as source coding, and the RCPC is used as channel coding.

Even though many JSCC methods have been introduced to improve the performance of 2D video transmission, very little work has been proposed for 3D video. An example of the JSCC for 3D video transmission application is proposed in [[xi]]. Unlike 2D video, two data streams, left and right views, needs to be considered for source coding of 3D content. This study considers the operation between H.264/AVC and rate compatible punctured turbo codes (RCPT) for source and channel coding respectively. To protect compressed video data from channel errors, the concept of unequal error protection (UEP) is employed to assign different levels of protection to each encoded data partition with regard to their decoding importance. The JSCC method introduced in this chapter aims to improve the performance of 3D video based on the colour-plus-depth 3D representation over the WiMAX based Reyleigh fading channel from a perceptual quality point of view.

1.2 Classical Rate-Distortion Theory

Rate distortion theory is a key aspect of information theory which provides the theoretical bounds for lossy data compression. Theories of rate distortion were created by Claude Elwood Shannon, known as “father of information theory”, in 1948 [[xii]], in his initial work on information theory. This addresses the problem of determining the minimal amount of information or entropy (R) that should be communicated over a channel, such that the source (input signal) can be approximately reconstructed at the receiver (output signal) without exceeding a given distortion (D). The relationship between rate and distortion is illustrated in 5‑1. Term rate is usually refers to the number of bits per data sample to be transmitted or stored and “distortion” refers to the degree of difference between original and reconstructed signals, usually evaluated by the mean squared error (MSE). However, as most lossy compression techniques operate on data samples that will be ultimately perceived by human consumers (e.g. watching video and pictures) the distortion measures should be intelligent to model on human perception. To date, since the human perception models are less well developed for image and video, lossy compression techniques still rely on simple statistical measure such as MSE, although with less correlation with regards to HVS.

Ideal noiseless transmission channels should produce the same massages or symbols, X, emitted by the source at the destination. However transmission impairments, such as noise, in the channel alter the emitted symbols, resulting in a different symbol space Y at the receiver. Consider a simple model of an error pone communicational channel shown in xxx. Let's assume XXXX illustrates the forward transitions of the channel. Here X and Y represent the alphabets of symbols transmitted and received respectively, during a unit time over the channel. Let, P(yj|xi) defines the conditional probability distribution functions of output symbols yj for a given input xi, where xiX and yj Y.

If the channel is intended to deliver y­j when xi transmitted, then the error probabilities are defined by P (yj|xi) for all j ≠ i. For the channel, mutual information, I(x­i;yj), measures the amount of information that symbols xi and yj convey about each other. I(x­i;yj), is defined as follows:


In practice, most transmission channels are consider to lie between perfect transfer (i.e. each yj uniquely identifies a particular xi) and zero transfer (i.e. yj is totally unrelated to xi ) Average mutual information is defined to analyse the general case.


H(X) denotes the entropy of the output signal X and H(X|Y) denotes the conditional entropy of the input signal (X) given the output signal (Y). Equation 5.2 states that the average information conveyed per symbol equals the source entropy minus conditional entropy.

The solution for rate and distortion problem can be achieved by minimising the rate-distortion function given below:


Where R is the information rate and D is an average distortion. I(X;Y) which describes the average mutual information between an original source (X : where the source selects symbols from an alphabet X) and a reconstructed data (Y),. Equation 5.1 says that for a given maximum average distortion Dmax, the rate distortion function R(D) defines the lower bound for the transmission bit-rate. The minimization is over all conditional probability distributions P (yj|xi) for which the joint distribution P (yj ; xi) satisfies the expected distortion constraint. The set of defines all the conditional distributions of P (yj|xi)

Conditional probability p(y | x) is considered as an inherent and fixed property of the communicational channel defined by the characteristics of the noise in the channel. The joint probability distribution of X and Y is entirely determined by the nature of the channel and the distribution of messages, f(x), to be transmitted over the channel. Under these constraints, the objective is to maximize the rate of information communicating over the noisy channel. The appropriate measure for this is known as the mutual information, The theoretical upper bound of mutual information is know as the channel capacity and is given by:

C = \max_{f} I(X;Y).\!

Channel capacity has the subsequent property related to transmitting information at rate R, where R is generally bits per message or symbol. For a communication system where the information rate R is < C and coding error ε is > 0, it is always possible to transmit with an arbitrarily small error, such that the maximal probability of error is less than an acceptable level ε. In addition, for any rate R > C, it is unachievable to transmit with arbitrarily small block error. The objective of channel coding is to find nearly optimal codes that can be used to transmit data over an error pone channels with an acceptable error at a rate close to channel capacity. However, in most practical video communication systems, the quality of transmitted video varies due to variations in the allowable bandwidth limitations. Thus, the maximum perceptual quality, under the rate constraint, can be achieved by the solving the following:


The set of Φ is defines the solution space of conditional distributions P (yj|xi) for which the joint distribution P (yj;xi) satisfies the expected rate constraint.

1.3 Joint Source and Channel Coding for 3D Video

In this section, the frame work of the proposed JSCC to improve the performance of colour-plus-depth 3D video transmission over wireless channels is discussed. The difference between the JSCC for 2D video and the JSCC for 3D video is that the traditional 2D video has only one source component while the 3D video consists of two source components: colour video and depth map.

The overall system model considered for the proposed JSCC for colour-and-depth representation of 3D video illustrated in 5‑3. At the transmitter (Tx), color and depth videos are separately compressed by H.264/AVC source coding and then protected by low-density parity-check (LDPC) codes. The output bit streams are rearranged to get single output at the multiplexer. Subsequently, the output from multiplexer is transmitted by WiMAX over a Rayleigh fading channel. At the receiver (Rx), received data stream is separated back to 2 data streams before decoded by LDPC and H.264/AVC decoders, respectively. At the end of the process, colour and depth map are reconstructed.

The main concept of JSCC is that both the source coding and channel coding are adapted according to channel conditions in order to minimize the distortion. Distortion in video communication can be separated into two major types. The first type is the quantization distortion introduced by lossy source encoding and the second type is caused by channel noise. These distortions are simply called “source distortion” and “channel distortion”. The overall distortion is equal to the addition of source and channel distortions. A popular measure for distortion is a mean square error (MSE). The overall picture distortion at the receiver end can be defined as the MSE between the received video frame and the original one. But it is a well known fact that due to the lack of correlations with the human visual system (HVS) MSE can not evaluate the quality as by a panel of human [13]. The main objective of this chapter is to investigate on minimizing the effect of these two types of distortions, using a JSCC approach, from a perceptual quality point of view.

H.264/AVC Source Coding

Since early 1990s, when the video coding technology was in its immaturity, international standards, sequentially, H.261 [[xiii]], MPEG-1 [[xiv]], MPEG-2/H.262 [[xv]], H.263 [[xvi]], and MPEG-4 (Part 2) [[xvii]] have been the motivation behind the success of digital video compression. In 2003 JVT (Joint video Team) developed H.264/AVC source coding standard and today it is considered as one of the most powerful video compression standards of all time. This can achieve almost twice the coding gain when compared to former video compression standards like H.263. Studies have shown that [[xviii]] H.264/AVC can not only offer high quality services for high-bandwidth networks but also an acceptable quality service for low-bandwidth services. H.264/AVC standard is capable of providing technical solutions to a broad rage of application areas that covers all varieties of digital compressed video, including video broadcasting, video on demand (VOD), Multimedia Messaging Services (MMS) and serial or interactive storage on magnetic and optical devices. Moreover, due to flexible and customizable deign of the codec new applications may be deployed over the existing architecture. This is because the design architecture covers a Video Coding Layer (VCL), which is intended to efficiently represent the video content, and a Network Abstraction Layer (NAL), which organize the VCL representation of the video in a manner to allow the same video syntax to be compatible in different network environments. These features, along with several others, aid H.264/AVC to perform considerably better than any prior video coding standard under a wide range of circumstances and application environments.

LDPC channel coding

LDPC codes were originally invented by Gallager in early 1960's [[xix]] as an error correcting code. After the invention LPDC codes were largely forgotten and was rediscovered by Mackay in 1999 [[xx]]. Since then they have experienced a remarkable return in the last few years. As a result of the significant development of the LDPC codes, it found an application as an optional channel coding technique to be used in the WiMAX standard [[xxi]]. LDPC codes are considered as capacity-approaching codes, which mean that the codes allow transmission at rates close to the theoretical maximum, as defined by the Shannon limit, for a symmetric memory-less channels over a very large code length. For instance, the performance of the LDPC code is only 0.0045 dB below the theoretical maximum, for a code length of one million bits.

LDPC codes in WiMAX is based on a set of (one or more) fundamental coding rates: 1/2, 2/3A, 2/3B, 3/4A, 3/4B, and 5/6. Each LDPC code in the standard is defined by a parity check matrix H of size m × n, where m refers to the number of parity bits and n refers to the length of output packet. The number of systematic bits (information bits) of the code is k = n - m. 2/3A, 2/3B and 3/4A, 3/4B has the same coding rates but different parity-check matrices H. Parity check matrix H is obtained by expanding the generator base matrix by replacing the entries with a square matrix (z × z), where the matrix dimension is equal to the block size. The base matrices defined in the IEEE 802.16e standard [69] for the six fundamental code rates are given in Appendix B.

Mobile WiMAX (Worldwide Interoperability for Microwave Access)

Mobile WiMAX is an emerging telecommunications technology for which provides mobile wireless internet access. The high data rate and Quality of Service (QoS) provided by WiMAX technology make it attractive to multimedia applications, such as video telephony, video gaming, and video broadcasting. IEEE 802.16e-2005 standard [75] [[xxii]] defines the formal specificities of Mobile WiMAX. Mobile WiMAX standard was developed by the WiMAX forum and is an amendment to IEEE 802.16d-2004, (Fixed WiMAX standard) to introduce support for mobility. The technology promises high data rates, up to 70 Mbps, and wide coverage, coverage radius of up to 50 km, at a lower cost. Consequently, Mobile WiMAX has gain the most commercial attention to date and is being successfully deployed in many countries. To accomplish high efficiency, throughput and reliability, several techniques are built into the MAC and Physical layers of the mobile WiMAX standard. In addition, security and quality of service (QoS) mechanisms are also incorporated. Packet structure of mobile WiMAX is well suited for non line of sight communication, which is the typical mobile WiMAX user experience. Fading distribution of non line of sight communication is closely correlated to Rayleigh distribution, because it processes the statistical time varying properties of non line of sight communication link [[xxiii]][[xxiv]][[xxv]]. Channel coding for error correction is necessary for reliable communication due to channel noise, fading and other transmission impairments. For this purpose the IEEE 802.16e-2005 standard suggests the use of following coding methods [[xxvi]]:

* Low-density parity check (LDPC) - allows six fundamental code rates: ½, 2/3A, 2/3B, 3/4A, 3/4B and 5/6

* Convolution Turbo Code (CTC) - allows fundamental code rate of 1/3 and additional code rates using puncturing: 2/3, ¾ and 5/6

* Convolutional Code (CC) - allows a mother code rate of ½ and additional rates using puncturing: 2/3, ¾ and 5/6

In most cases LDPC codes is preferred for channel coding in mobile WiMAX. Advantages of LDPC over CTC and CC codes are discussed in [80].

[i][ ] S. Zahir Azami, P. Duhamel, and O. Rioul. Joint source-channel coding: Panorama of methods. CNES Workshop on Data Compression, Toulouse, France, November 1996. http://citeseer.nj.nec.com/zahirazami96joint.html.

[ii][] M.J. Ruf, and J.W. Modestino, “Operational rate-distortion performance for joint source and channel coding of images,” IEEE Trans. on Image Processing, vol.8, no.3, pp. 305-320, 1999.

[iii][] J. Song, and K.J.R. Liu, “Robust progressive image transmission over OFDM systems using space-time block code,” IEEE Trans. on Multimedia, vol.4, no.3, pp. 394-406, 2002.

[iv][] R. Hamzaoui, V. Stankovic, and X. Zixiang, “Optimized error protection of scalable image bit streams [advances in joint source-channel coding for images],” IEEE Signal Processing Magazine, vol.22, no.6, pp. 91-107, 2005.

[v][] M.F. Sabir, H.R. Sheikh, R.W. Heath, Jr., and A.C. Bovik, “A joint source-channel distortion model for JPEG compressed images,” IEEE Trans. on Image Processing, vol.15, no.6, pp. 1349-1364, 2006.

[vi][] S. Zhao, Z. Xiong, and X. Wang, “Optimal resource allocation for wireless video over CDMA networks,” IEEE Trans. on Mobile Computing, vol.4, no.1, pp. 56-67, 2005.

[vii][] J. Wang, M. Venkatachalam, and Y. Fang, “System Architecture and Cross-Layer Optimization of Video Broadcast over WiMAX,” IEEE Journal on Selected Areas in Communications, vol.25, no.4, pp. 712-721, 2007.

[viii][] M. Stoufs, A. Munteanu, J. Cornelis, and P. Schelkens, “Scalable Joint Source-Channel Coding for the Scalable Extension of H.264/AVC,” IEEE Trans. on Circuits and Systems for Video Technology, vol.18, no.12, pp. 1657-1670, 2008.

[ix][] F. Zhai, Y. Eisenberg, T.N. Pappas, R. Berry, and A.K. Katsaggelos, “Rate-distortion optimized hybrid error control for real-time packetized video transmission,” IEEE Trans. on Image Processing, vol.15, no.1, pp. 40-53, 2006.

[x][] A.J Goldsmith, and M. Effros, “Joint design of fixed-rate source codes and multiresolution channel codes,” IEEE Trans. on Communications, vol.46, no.10, pp. 1301-1312, 1998.

[xi][] P.Y. Yip, A. Malcolm, W.A.C. Fernando, K.K. Loo, and H.K. Arachchi, “Joint source and channel coding for H.264 compliant stereoscopic video transmission,” Canadian Conf. on Electrical and Computer Engineering, pp. 188-191, 2005.

[xii][] The Bell System Technical Journal,Vol. 27, pp. 379-423, 623-656, July, October, 1948.

[xiii][] ITU-T, “Video codec for audiovisual services at px64 kbits/s,” ITU-T Rec. H.261 v1: Nov 1990, v2: Mar. 1993.

[xiv][] ISO/IEC JTC 1, “Coding of moving pictures and associated audio for digital storage media at up to about 1.5 Mbit/s - Part 2: Video,” ISO/IEC 11172 (MPEG-1), Nov. 1993.

[xv][] ITU-T and ISO/IEC JTC 1, “Generic coding of moving pictures and associated audio information - Part 2: Video,” ITU-T Rec. H.262 and ISO/IEC 13818-2 (MPEG-2), Nov. 1994 (with several subsequent amendments and corrigenda).

[xvi][] ITU-T, “Video coding for low bit rate communication,” ITU-T Rec. H.263; v1: Nov. 1995, v2: Jan. 1998, v3: Nov. 2000.

[xvii][] ISO/IEC JTC 1, “Coding of audio-visual objects - Part 2: Visual,” ISO/IEC 14496-2 (MPEG-4 Part 2), Jan. 1999 (with several subsequent amendments and corrigenda).

[xviii][] Available online at http://nccur.lib.nccu.edu.tw/bitstream/140.119/32688/5/303605.pdf

[xix][] Gallager, R. G., “Low Density Parity Check Codes”, PhD dissartation, M.I.T. Press, Cambridge MA 1963.

[xx][] D. MacKey, “Good error correcting codes based on very sparse matrices” IEEE Transactions on Information Theory, vol 45, ni. 2, pp. 399-431, March 1999.

[xxi][] IEEE P802.16e/D12, “Draft IEEE standard for local and metropolitan area networks part 16: air interface for fixed and mobile broadband wireless access systems,” 2005.

[xxii][] IEEE 802.16. Worldwide Interoperability for Microwave Access, www.wimaxforum.org.

[xxiii][] Salmon B. P., “Optimizing LDPC codes for a mobile WiMAX sustem with a saturated transmission amplifier”, Master's dissertation, University of Pretoria, South Africa, 2008.

[xxiv][] J. Proakis and M Salehi, Communication system engineering, 2nd ed. Upper Saddle River: Prentice Hall, 2002

[xxv][] L. Staphorst, “Viterbi decoded linear block codes for narrowband and wideband wireless communication over mobile fading channels”, Master's thesis, University of Pretoria, South Africa, 2005.

[xxvi][] TurboBest group “IEEE 802.16e LDPC Encoder/Decoder Core”, White paper, available at http://www.turbobest.com.