Power consumption estimation for H.264/AVC and DVC for wireless networks

Introduction

Wireless sensor networks (WSN) are rapidly emerging as a framework to carry out distributed and pervasive applications such as environmental monitoring, habitat studies, and video-surveillance, just to mention a few. Such networks typically consist of tens to hundreds of small-size and low-cost nodes. Each node is equipped with a sensing device that collects information from the environment (e.g., temperature, vibrations, images), and transmits it through the network to a gateway node. Wireless video-surveillance networks (WVN) are gaining increasing popularity due to the number of applications they make possible to carry out.

So, during the last years of video codec development more and more attention has been paid to the low-complexity codecs, as they are considered now to be used in these wireless sensor networks and another systems, where it is necessary to decrease encoding power consumption so that they can achieve longer working time, and the decoder power consumption for this system is not an important issue. Energy consumption is a critical aspect, even more than in other sensor networks since video cameras collect a huge amount of data that must be transmitted over the wireless link. Therefore, power consumption on the encoder side became one of the most important issues along with compression efficiency.

Project description

There are several ways used by different teams for approaches for comparison. For example, Dragotti and Gastpar in their book about DVC [1] used processing time measurements. They have defined the working station, set up encoding parameters and run two codecs – DISCOVER project[2] from DVC side and JM reference software[3] from H.264 and measured the processing time. However, this approach has following disadvantage. These programs were not optimized and therefore this estimation can be considered preliminary. Another way of complexity comparison consists of dividing algorithm into simple functions and calculating simple operations like shifting, addition and multiplication. This approach was used for instance for calculating DCT complexity by William B. Pennebaker, Joan L. Mitchell in their book[4]. However, most effective results are obtained by using real power consumption estimation and analysis, like it was done in the following work [5]. In our work we tried to combine these approaches and firstly divided two algorithms into different steps and then tried to calculate relative power consumption [6].

The second stage of the project takes into account also the power needed for transmission of the data from the encoder. In principle, video compression can reduce the amount of data to be transmitted by a considerable factor. On the other hand, it is well-known that most video coders exhibit a very high computational burden. This is not a matter of concern in desktop multimedia applications, in which one can afford a 2 GHz processor to encode and decode video at 30 frames per second in real-time. However, the use of such a powerful encoder in a WVN application may be objectionable, in that it is possible that the energy saved by transmitting less data does not compensate for the energy required to compress the video data. Therefore, the trade-off between communication and computation is a crucial aspect that need to be investigated [7].

Project goals and future research directions

The aim of this work is to show that although DVC is considered to be low-power approach for video encoding [8], common approach based on H.264/AVC in differential frame coding mode can achieve a comparable performance in complexity and power consumption. The second phase of the project refers to the power consumption estimation taking into account transmitter power consumption.

Project timeline and expected deliverables

• October 2009 –project started
• February 2010 – final results for comparison received
• April 2010, 7th FRUCT seminar – project presentation
• May 2010 – paper submission (about encoder power consumption)
• June 2010 – Second stage of the project
• September 2010, SoftCOM conference - first paper presentation (about encoder power consumption)
• November 2010, 8th FRUCT seminar – project presentation
• December 2010 – second paper submission (about overall power consumption for encoding and transmission)

Call for participation

We would like to invite all of you interested in this project to take part in the research in this area. Any comments are welcome by the contact e-mail.

List of team members

Ann Ukhanova, Ph.D student, Technical University of Denmark

Eugeniy Belyaev, Ph.D., SPIIRAS

Søren Forchhammer, professor, Technical University of Denmark

Yacine Ghamri-Doudane, Ph.D, Associate Professor, University Paris-Est-Marne-la-Vallée

Ismail Salhi, Ph.D student, University Paris-Est-Marne-la-Vallée
 

Contact details

Ann Ukhanova - ann...@fotonik.dtu.dk

References

1. P.L. Dragotti, M. Gastpar Distributive Source Coding. Theory, Algorithms, and Applications, Elsevier, 2009
2. DISCOVER codec, available on: http://discoverdvc.org/
3. H.264/AVC JM Reference Software, available on: http://iphome.hhi.de/
4. William B. Pennebaker, Joan L. Mitchell, Jpeg: Still Image Data Compression Standard, 1992
5.  Z. He, Y. Liang, L. Chen, I. Ahmad, D. Wu, Power-Rate-Distortion Analysis for Wireless Video Communication Under Energy Constraints, IEEE Transactions on Circuits and Systems for Video Technology, Vol. 15, No. 5, 2005
6. A. Ukhanova, E. Belyaev, S. Forchhammer, Encoder power consumption comparison of Distributed Video Codec and H.264/AVC in low-complexity mode, accepted for SoftCOM conference, Croatia, September 2010.
7. C. Chiasserini ,  E. Magli, Energy Consumption and Image Quality in Wireless Video-Surveillance Networks, Proceedings of the 13th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications, 2002
8. A. Aaron and B. Girod, Wyner-Ziv video coding with low-encoder complexity, Proc. Picture Coding Symposium, PCS, San Francisco, CA, December 2004. Invited paper.