Evaluation of Video Transmission using CSVC (Combined Scalable Video Coding)
The uses of streaming video application on wireless broadband network services have significant development. Those require more bandwidth significantly. On the other side, the use of wireless technologies as increasing demand of mobile multimedia services make the important use of fluctuate bandwidth. Wireless media communication has low reliability with bandwidth fluctuations that lead to degradation video quality significantly. One attractive method to overcome those problems is by scaling bit stream into a number of scales of priority (layer), in the form of base-layer and enhancement layer, which is known as scalable video coding (SVC) method [1].
Some researches on the vision of SVC over broadband services are reviewed in [2], while the performance of SVC and its applications are shown in [3]. The combined scalable video coding (CSVC) was proposed by various studies such as hybrid temporal-FGS (fine granular scalability) by Mihaela and spatio-temporal by Blaszak. CSVC adopted by the joint scalable video model (JSVM) proposed by Schwarz [4]. The recent applications of SVC on broadband service are proposed in Maodong Li. There are two schemes proposed for the SVC as a public evaluation platform based on open-source program, that is SVEF (scalable video coding evaluation framework) by Andrea Detti and EvalSVC (evaluation of SVC) by Tien Anh Le. Current researches are still using SVC with JSVM in general approach (e.g., only using one mode scalable) but the characters of the input (source sequences) and the transmission channel will influence of performance end-to-end system. It is reason why in this research using combination of SVC. We hope will make it more flexible for adaptation to input characters, fluctuation of channel transmission, and multicast of network conditions.
In this works, we use dominant CSVC to exploit the SNR(Signal to Noise Ratio)-scalable, especially in the medium grain scalable (MGS) and the coarse grain scalable (CGS) modes. To the best of our knowledge, CSVC scheme using the MGS and CGS to enhance the quality of video streaming over broadband wireless network has not been well studied yet. This research is focuses more to comparison of the MGS to CGS as the development of our previous work [5], [6].
A. Scalable Video Coding (SVC)
SVC is part of standard development of H.264/MPEG-4 part 10 AVC (Advanced Video Coding), or H.264/AVC. The development processes took a long 20 years, start from H.262 and MPEG-2, followed by H.263+ and MPEG-4. Until now, SVC standard is still in amendment, and is a cooperative work of many parties to establish JSVM standard [7]. Since January 2005, MPEG and VCEG join in JVT to carry into completion the amendment H.264/AVC as an official standard.
Scalability proposed for the first time to reduce packet (cell) loss in ATM networks. It created 2 (two) groups of bit stream or layer, base-layer and enhancement-layer. Layer containing vital information is base-layer, while enhancement–layer loaded by residual information to improve image quality or video to be produced, as shown in Fig.1.
Fig.1. The idea of Scalability Coding System
Within data transmission process, in case of congestion in transmission channel, at least base-layer containing vital information still make it through. There are three types of scalability method as in Fig.1: SNR (Signal to Noise Ratio) Scalability, Spatial Scalability, and Temporal Scalability. Beyond that, there exists one type of scalability that combines the first three.
B. Combined Scalability Video coding (CSVC)
The implementation of combined scalable is based on structure and efficiency of SVC coding. In case that the inter layer resolution changed, spatial scalable will be dominant. SNR scalable is dominant in changes of SNR, and temporal scalability dominant in changes of rate. Combination of those three is likely to be because of varying in sequences characteristics, fluctuated network condition and multiterminals [4]. This research utilizes three layers combined scalable, which includes 1 base layer and 2 enhancement layers, as shown in Fig.1. Block diagram of encoder-decoder with 3 layers combined scalable shown in Fig. 2.
Fig. 2. Structure of Encoder 3 layer SVC
C. MGS and CGS
Each spatial layer is referred to as CGS or MGS. CGS can be viewed as a special case of spatial scalability in H.264 SVC, in that similar encoding mechanisms are employed but the spatial resolution is kept constant. More specially, similar to spatial scalability, CGS employs inter-layer prediction mechanisms, such as prediction of macro block modes and associated motion parameters and prediction of the residue signal. CGS differs from spatial scalability in that the up-sampling operations are not performed. In CGS, the residual texture signal in the enhancement layer is re-quantized with a quantization step size that is smaller than the quantization step size of the preceding CGS layer. SVC supports up to eight CGS layers, corresponding to eight quality extraction points, i.e., one base layer and up to seven enhancement layers. When quality scalability is using CGS, switching between different CGS layers must be at the defined point but the quality scalability is using MGS, switching different MGS layer can be in any access unit. With the MGS concept, any enhancement layer NAL unit can be discarded from a quality scalable bit stream. For the first quality layer is restricted to CGS, and enhancement layer is consist of fine grain scalability (FGS).
D. Evaluation Framework
The scheme in this research is relatively new [6], where the use of CSVC application that implemented on wireless broadband network as WLAN IEEE 802.11e on Network Simulator II (NS2) [6]. Fig. 3 shows research scheme using input BUS.cif video sequence. This research proposes the use of CGS and MGS modes of JSVM. This experiment uses JSVM version 9.19 [7] and NS2 version 2.29 [8].
Fig. 3. The scheme of research using JSVM with NS-2
We presented scheme for platform evaluation of video transmission based on the CSVC over broadband wireless network (WLAN IEEE 802.11e), simulated by NS-2. We also investigated the impacts of the use of MGS and CGS modes on performance of this system. The applications of MGS mode on CSVC increases the performance compared to CGS mode. In general, our scheme is implementable in video sequence on broadband wireless service.
References:
6. K. Rantelobo, Wirawan, G. Hendrantoro, A. Affandi, and H. Zhao, ”A New Scheme for Evaluating Video Transmission Over Broadband Wireless Network”, Future Wireless Networks and Information Systems: Lecture Notes in Electrical Engineering, Series Vol. 143, Springer Berlin Heidelberg, pp. 335-341, 2012. (download pdf)
7. ITU-T and ISO/IEC (JVT): Joint Scalable Video Model JSVM version 9.19. March, 2011. [Online]. Available: http://wftp3.itu.int/av-arch/jvt-site/
8. The Network Simulator II (NS-2) [Online]. Available at: http://www.isi.edu/nsnam/ns/
Additional References:
a. Richardson, Iain E. G, The H.264 advanced video compression standard – 2nd ed., John Wiley & Sons, Ltd., 2010.
c. Overview SVC: Extension of H264-AVC (from web: Fraunhofer Heinrich Hertz Institute (HHI))
d. Scalable Video Coding on Wikipedia
e. SVC on MPEG