Private:mobileTV

Mobile TV Networks

Mobile TV allows users to watch their favorite TV shows and games on small hand-held devices while traveling. It, therefore, extends the Prime Time viewing of users and provides more business opportunities for content providers. The market for mobile TV is huge: it is expected to grow to up to 20 billion Euros with 500 million customers by 2011[1]. In fact, mobile TV has already been deployed in parts of Europe and Asia and in pilot-testing in several locations in North and South Americas (official DVB-H site). This rapid adoption is fueled by the desire of users for multimedia content and by the technological advances in wireless mobile devices, such as personal digital assistants (PDAs), smart cellular phones, and mobile media players. Many of these devices have evolved to almost full-fledged mobile computers with high resolution displays, fast network links, large memory and storage space, and fast processors. Therefore, multimedia content can be rendered on most of these mobile devices, which further stimulates the user demands for more content and better quality.

We consider various quality-of-service metrics and propose efficient algorithms to maximize them in mobile TV networks. The considered metrics include: energy saving, channel switching delay, and network utilization. For mobile TV users, energy saving and channel switching delay are the two most important metrics. This is because higher energy saving results in longer watch time, and longer channel switching delay degrades view experience as many users quickly flip through numerous TV channels before they decide to watch the specific ones. For mobile TV network operators, network utilization is the most important issue, because wireless spectrum is very expensive: often costs multi-million dollars. We have proposed several algorithms to: (i) maximize energy saving on mobile devices, (ii) guarantee the switching delay from a TV channel to any other channel is short, and (iii) maximize the number of channels concurrently broadcast within a given wireless spectrum. We analytically analyze the proposed algorithms and conduct extensive simulations to evaluate their performance. Most importantly, we have also implemented a real mobile TV testbed in our Lab, which provides us a realistic platform for analyzing the performance of the proposed broadcast schemes. The mobile TV testbed has two parts: a commodity Linux box as the base station and several smart phones as receivers. Our simulation and experimental results indicate that the proposed broadcast schemes are: (i) optimal/near-optimal, (ii) efficient and scalable, and (iii) practical for real mobile TV networks.

People

• Mohamed Hefeeda

• ChengHsin Hsu (PhD student)

• Yi Liu (MSc student)

• Cong Ly (MSc student)

Publications

• C. Hsu and M. Hefeeda, Broadcasting Video Streams Encoded with Arbitrary Bit Rates in Energy-Constrained Mobile TV Networks, IEEE/ACM Transactions on Networking, Accepted August 2009.

• M. Hefeeda and C. Hsu, On Burst Transmission Scheduling in Mobile TV Broadcast Networks, IEEE/ACM Transactions on Networking, Accepted July 2009.

• C. Hsu and M. Hefeeda, On Statistical Multiplexing of Variable-Bit-Rate Video Streams in Mobile Systems, In Proc. of ACM Multimedia 2009, Beijing, China, October 2009. Acceptance: 18%.

• Y. Liu, C. Hsu, and M. Hefeeda, On the Benefits of Cooperative Video Broadcast over WMANs and WLANs, In Proc. of ACM Multimedia 2009, short paper, Beijing, China, October 2009. Acceptance: 30%.

• C. Hsu and M. Hefeeda, Multi-Layer Video Broadcasting with Low Channel Switching Delays, In Proc. of IEEE International Packet Video Workshop (PV'09), Seattle, WA, May 2009.

• C. Hsu and M. Hefeeda, Video Broadcasting to Heterogeneous Mobile Devices, In Proc. of IFIP Networking 2009, Aachen, Germany, May 2009. Published in Springer-Verlag Lecture Notes in Computer Science, LNCS 5550, pp. 600--613, 2009. Acceptance: 20%

• C. Hsu and M. Hefeeda, Time Slicing in Mobile TV Broadcast Networks with Arbitrary Channel Bit Rates, In Proc. of IEEE INFOCOM 2009, pp. 2231--2239, Rio de Janeiro, Brazil, April 2009. Acceptance: 20%

• C. Hsu and M. Hefeeda, Bounding Switching Delay in Mobile TV Broadcast Networks, In Proc. of ACM/SPIE Multimedia Computing and Networking Conference (MMCN'09), San Jose, CA, January 2009.

• M. Hefeeda, C. Hsu, and Y. Liu, Testbed and Experiments for Mobile TV (DVB-H) Networks, ACM Multimedia'08 Technical Demonstration, Vancouver, Canada, October 2008. Best Demo Award

• M. Hefeeda and C. Hsu, Energy Optimization in Mobile TV Broadcast Networks, In Proc. of IEEE International Conference on Innovations in Information Technology (Innovations'08), pp. 430--434, Al Ain, United Arab Emirates, December 2008. Best Paper Award.

Press Coverage

• July 1, 2009: Our mobile TV research is also featured in the July issue of the ACM Tech News: see article or local PDF

• June 26, 2009: Our mobile TV project is featured on CTV News: see article or local PDF

• June 15, 2009: Omni-TV featured Cheng and our mobile TV project (in Mandarin): local mpeg file

• June 4, 2009: The mobile TV project is in World Journal News (in Mandarin): see article or local PDF

• May 28, 2009: The mobile TV project is in SFU News: see article or local PDF

Mobile TV (DVB-H) Testbed

We have implemented a complete end-to-end testbed for DVB-H (Digital Video Broadcast--Handheld) networks. The testbed provides a realistic platform for analyzing various aspects of these networks, including the energy saving achieved by the time slicing mechanism, average channel switching delay, network capacity in terms of number of TV channels that can be broadcast, visual quality of TV channels transmitting different types of video streams, information exchange and interactivity between base station and receivers, among many others. To the best of our knowledge, there exists no complete open-source testbed for DVB-H. The details of testbeds and pilot networks created by companies are usually not published, and the source code is not available. Thus academic researchers designing algorithms and protocols for mobile TV networks, including ourselves, had to resort to simulation and/or theoretical analysis. To address this problem, we make the details and source code of our testbed available to the research community.

The main components of our mobile TV testbed are shown in the following figure.

Base Station. The base station is a Linux box (Intel Quad-Core Xeon E5420 (2.5 GHz) PC running Ubuntu Linux) in which we installed the RF signal modulator: Dektec DTA-110T DVB-T/H Modulator and UHF Upconverter for PCI Bus. This modulator implements the physical layer of the protocol stack and transmits DVB-H standard compliant signals via an indoor antenna. The RF output level of the modulator, however, is quite low (-29 dBm) and can only reach up to 1-meter broadcast range with a 6 dB receiver antenna. Using a low-power amplifier, the RF signal can be boosted to about 0 dBm, which gives us approximately 20-meter range in our lab environment.

IP packets of the video streams are encapsulated in MPEs and FEC-coded using an IP Encapsulator. In the initial setup, we adopt an open-source IP Encapsulator, which implements simple time slicing techniques. We extended that IP encapsulator to support more sophisticated and optimal time slicing schemes. Recently, we have re-designed the base station software to be well-structured with defined interfaces in order to facilitate implementing and comparing different current/future algorithms, including multimedia streaming and time slicing algorithms. This new base station design follows multi-threaded paradigm, and can fully utilize multi-processor systems, which is common nowadays. We continue improving the base station implementation: the latest source code can be browsed at the subversion server.

Receivers. We use the Nokia N92 and N96 device as receivers. These devices are equipped with the receiver-side of the DVB-H protocol and video player. The operating system on this device (Symbian) provides several APIs, including APIs for measuring energy consumption. While the N92 device helps in assessing the visual quality of videos, it does not provide detailed logging functions of the low-level signals, which are needed to evaluate the performance of different protocols. To address this shortcoming, we added the Divi Catch RF-T/H transport stream analyzer to the testbed. This analyzer can be attached to a PC via a USB port. The analyzer records traffic streams as well as provides a very detailed information on the RF signal, the MPEs, jitter, time slicing, and so on. It also comes with a visualization software that can run on the PC for analysis.

Software

We have two testbed implementations: (i) FATCAPS base station, and (ii) integrated base station. The former one is based on FATCAPS implementation, while the later one is implemented by us from scratch.

FATCAPS Based Implementation

We initially implemented our testbed on top of FATCAPS, which is an open source time slicer implementation. Several documents are listed in the following.

• Testcases abstracted in early 2008
• Setting up a base station using FATCAPS
• Overview of PSI/SI
• Overview of Electronic Service Guide (ESG)
• Overview of FLUTE/ALC protocol
• Juice: an energy profiling tool
• We have collected several MPEG-2 TS stream traces. These streams are compatible with recent Nokia handsets, such as N96. (We thank Abertis Telecom for providing us some of these traces. )
  1. Nokia stream #1
  2. Nokia stream #2
  3. Nokia stream #3
• Implementing ESG

MTV Implementation

We have implemented the software of DVB-H base stations as an open-source project called mtv. Unlike commercial products, mtv allows researchers to implement their ideas and algorithms on top of it. Thus researchers can evaluate the real performance of their ideas in an open-source experimentation platform that is very close to deployed networks.

The latest base station code can be browsed a the subversion server. We continue improving the base station implementation, and we list ongoing/future works in this document.

Discussion and Ideas

• Meeting Minutes (login required)

• Underinvestigated ideas of streaming to mobile devices (login required)

References and Links

• Q. Gao, M. Chari, A. Chen, F. Ling, and K. Walker, "MediaFLO Technology: FLO Air Interface Overview": This book chapter explains why mediaFLO achieves short channel switching delay. MediaFLO employs 1-sec superframe and each MLC is transmitted as four bursts in every superframe. Therefore, when user switches the channel in superframe x, the receiver will get all the four bursts in superframe x+1, which takes 1 to 2 secs, where 1 is the best case and 2 is the worst case. This article also indicates that the time-frequency assignment need not be rectangle, which allows finer grained resource allocation and thus better statistical multiplexing gain. However, this paper does not address the relationship between energy saving and time-freq allocation. There is only one sentence saying that the receiving circuit should avoid modulating the symbols irrelevant to the current channel to save energy.

• M. Chari, F. Ling, A. Mantravadi, R. Krishnamoorthi, R. Vijayan, G. Walker, and R. Chandhok, “FLO physical layer: An overview,” IEEE Transactions on Broadcasting, vol. 53, no. 1, pp. 145–160, March 2007.

• Media FLO Technology

• ATSC Mobile DTV

• The official Mobile TV website maintained by the DVB Project Office.

• Wing-TV Project: Docs to test and to verify in detail the DVB-H specification.

• J. Wang, M. Venkatachalam, and Y. Fang, System Architecture and Cross-Layer Optimization of Video Broadcast over WiMAX, IEEE JSAC 25(4), pp. 712--721, MAY 2007. (Check this special issue of JSAC as well.)

• G. Gardikis, G. Xilouris, C. Skianis, Broadband multimedia on the move with DVB-H, Multimedia Tools and Applications, 36(1-2), January 2008. (DVB-H tutorial in inter-active mode.)

• LIMBOS LInux setup for accessing Mobile Broadcasting through Online Streaming