Wave Report

This is a small post CES conference run by the IEEE’s Consumer Electronics Society. It is technical with a strong emphasis on research. Virtually all the audience and presenters are from Asia or Europe, with the vast majority from Asia. The US participation in CE is but a shell of its former leadership position

0803.1 Intel Defines its View of the Future of CE

Bradley Daniels, VP, Consumer Electronics Group, Intel, gave the talk: “Meeting the Challenges of Digital Convergence on CE Platforms.”

     CE is at an inflection point. The analogy was used with the GE-90 aircraft engine. There was a fundamental change in
     the need for long range aircraft, GE saw this, created a new engine for the emerging market and, as a result, is
     dominating the market for engines on these transports. A similar inflection point is happening in CE and it is time
     to think what is the technology which will capitalize on this inflection? Andy Grove has spoken of the opportunity
     which inflections provide to create new markets and grow companies. Buy inference Intel is responding to that
     inflection in CE, as presented today.

     CE has been focused on the television in the home. The history of the television illustrates the inflection which
     is upon us. The first era was called CE 1.0 and it was analog devices in an analog world. The TV set, circa 1970
     was shown. The next transition, CE 2.0, is about digital devices in a digital world. Shown was a large screen LCD
     panel with a television image. The next transition, CE 3.0, and the inflection point, is when the “…TV Experience (is)
     Internet Compatible.” The importance of this is the market research that 61% would like to connect their TV to the
     Internet. In Intel’s view “The Internet Changes the Game.”

     Supporting the role of the Internet is that 16% of US households watch TV broadcasts online. It is time for the
     Internet to go from the den to the living room. Another transition is that 42% of the broadcast prime time viewing
     occurs through some form of DVR. The user experience in CE 3.0 is that television is based on choice, what one wants
     and when they want it.

Some of the characteristics of CE 3.0 are:

          Video streams
          Image quality
          Audio
          Graphics and
          Easy to use Navigation

All of these create significant demands on computation requirements and memory bandwidth. A chart was shown with
the impacts of the video codec on memory bandwidth.

Another function which demands processing resources is display processing. This was placed into three categories:

          Correction
          Enhancements and
          Conversion

     Intel stated that it is critical that we are moving into an HD world and a lot of content is not HD. We, as suppliers
     of technology, must do the necessary processing and conversion to make all video viewing as close to HD quality
     as possible. This takes processing power and memory bandwidth.

     This rich television environment will need better navigation. It was cited that we can learn much from the OS
     UI developments and integrate 3D navigation into the interface – the role of Microsoft Vista was implied. A side
     by side 2D and 3D navigation example was shown. Apple TV was another example shown on how the UI will change to make
     television easier to control.

New business models will be created. Advertising will be huge.

Audio was cited as a critical part of the CE 3.0 experience. Again, the processing requirements for audio
were cited including encoding and re-encoding.

   The next generation of CE devices driven by CE 3.0 will be complex and demanding. To further support this it was
     implied that the memory bandwidth requirement for display processing is significant (the actual numbers were left off
     the chart and table). What is needed is a “common platform.”

     To meet all these demands Intel is building a “IA System on a Chip.” This has a core processor, memory controller, I/O,
     audio/video pipeline, graphics and display processing and security. This will be a powerful platform for innovation
     services, according to Bradley.

When asked about software Bradley was emphatic that it would Linux. Intel would be supplying APIs. More
announcements would be forthcoming in the next two quarters.

0803.2 Consumer Electronics Technologies Developments

Technologies addressed included: content, storage, electronics, networks, servers, devices or products and human interface.

0803.2.1 Content

0803.2.1.1 Content Recommendation

The University of Vigo, Spain developed a technique which overcomes one of the shortfalls of content-based filtering based recommendation engines – overspecialization. The examples cited are of television program selection. The technique uses semantic reasoning which develops semantic relationships with a domain ontology and user modeling. This is then processed using spreading activation which explores relationships among nodes in a network where the network is the domain ontology. The authors cite significant improvements, up to 46%, over other recommendation engines.

0803.2.1.2 Content Analysis

Philips Research has developed a technology which automatically detects highlights in a soccer game. The motivation is to highlight, on mobile devices for example, activities or sequences that would interest a fan. The technique is based on audio frequency bands based on the audience response during the game. The frequency band information is used to detect periods of interest and the frequency content is the basis for scene extraction. One of the measures is the ability to detect goals. Goal recall was found to be 80% to 100%. It was stated that this technique could be applied to other sports.

0803.2.1.3 Content Preservation

Brigham Young University described their efforts to create recording materials for CD and DVDs which will last 1000 years. Cited was data that current optical media will only last 7 – 25 years. The presentation only implied that the use of such materials as gold, stone and glass could be used but no details were provided.

0803.2.2 Storage - Optical Media

Mitsubishi Electric discussed signal processing techniques to implement optical media which is grater than 25GB. This is a refinement on the Partial Response Maximum Likelihood (PRML) method to improve Inter-symbol Interference (IS)I. Experiments were run on a disk of 30GB density and BER of 1E-05 was achieved. This was claimed with using most of the existing PRML technique.

4GOOD is a European project with German and French support. The objective is to raise the storage density of optical media 3X to 4X than of Blu-ray. All aspects of the optical storage chain are being addressed and include: optics, media and electronics. It is expected that disk and drive demonstrators will be shown in the spring of 2008.

Hitachi has developed an optical pickup for a video camera drive which is compatible with both Blu-ray and DVD. This uses a dual lens method, a new collimator lens and a front monitor for both Blu-ray and DVD. The optical train to implement the slim pickup is complex.

Philips is looking to provide 3D for the mass market using the Blu-ray disk. It will be necessary to provide 2 HD decoders to implement the stereo and possibly 2 more SD decoders. The next consideration is the internface. HDMI 1.3 as an option supports deep color modes which can be adapted to stereo. Further, this same spec supports dual images. Thus, it is expected that 3D display of content could be backward compatible with the existing formats and specifications.

0803.2.3 Electronics

0803.2.3.1 Radio

Harman/Becker Automotive Systems is developing a software defined radio which supports AM/FM and the weather band. The following was implemented in software: FM/AM demodulation, quality estimation, interference cancellation, optional diversity, stereo demodulation, programmable bandwidth filers, RDS demodulation, audio blending and audio processing. It is claimed that the implemented radio has higher performance than a conventional design.

0803.2.3.2 Imaging

National Taiwan Normal University has developed processes to fuse multiple images in high dynamic range visual conditions. The objective is to show images which represent these scenes on conventional displays. The technique uses edge detection and maps to determine the luminance levels. The net image dynamic range is extended by 256X.

Korea University has developed a multi-image technique to estimate motion created blur and create a fused image which minimizes the effect of the blur. The technique uses 3 exposures – 2 long with a short one in between – to estimate the motion and compensate. The technique is able to lessen the impacts of the motion blur but cannot preserve the original color of the image, because this is based on the short time exposure.

Chung-Ang University, Korea, has developed a means for creating panorama images from multiple overlapped images in real time with a mobile camera. The technique operates in a semi-automatic and automatic modes. Images with a field of view to 180 deg were produced. The panorama composition time was 1.5 seconds.

Seoul National University and Samsung has developed a real time panorama mosaic system for a digital camera. This was accomplished on a PCA phone. It was possible to make a panorama from 4 640X 480 images in 1.2 sec from 4 photographs.

0803.2.4 Audio

NTT Corp, examined how an array microphone could be used for a home HD video conference product/service, which they call HDVS – HD video phone system. The application is similar to a Cisco Telepresence system but for the home and based on a home large LCD panel. What NTT sought to do is create a stereo presence of the participants of the conference. The stereo microphone based, on an array microphone, they developed did support the quality effect they sought to create.

Sony built a directional speaker system based on the nonlinear theory of air. This is based on generation of ultrasonic sound which is transmitted in the air. The interaction of the ultrasound with the nonlinear characteristics of the air will generate the frequency components which are audible. It is then possible to create highly directional sound. The Sony research addresses three issues with this approach: high harmonic and non-harmonic distortions and -120db/octave dependence for the low frequency response. Using modulation schemes it is possible to reduce the harmonic distortion. The bandwidth of the system can also be enlarged by selection a proper equalization and part of the the transduced array for each frequency band. Sony reports greatly improved sound quality.

0803.2.5 Video

Philips examined how PC PGUs could be used for video processing. This study GPUs for video motion estimation, executing the 3D Recursive Search Block Matching (3DRS) motion estimator. Algorithms were implemented in Direct3D and HLSL. Using the nVida 8800GPU it was shown that 49frames/second could be supported at 1920 X 1080p.

Fudan University examined the implementation of VCR functionality on a P2P VoD system. The key issues are high bandwidth consumption and response delay. Video segmentation is used based on a shot model that calculates the difference in adjacent frames base on a 192 dimension histogram in hue saturation. This system generates a frame media file and indexing file for every movie after the shot and scene segmentation. When a peer selects a movie the indexing file is sent. User only need to select on one shot or scene to select the content. The process was simulated and show reduction in bandwidth and improve the user experience.

0803.2.6 Video Camera

Hitachi has developed a camera based on Blu-ray recording. This is a small hand held video camera. On a 8cm disk it is possible to record 1 hour of HD video which is 7.5GB of data. The camera uses a 5Mp CMOS sensor. Sanyo and Chiba University studied the form factor for a hand held video camera which reduced fatigue in holding. Using electronmyograms (EMGs) to measure muscle load response a number of camera designs were tested. The variable was the grip angle of the means of holding the camera. The result was a grip of 105 degrees. This has resulted in a ultra compact full HD video camera.

Hitachi has developed a hybrid HDD/DVD camcorder. This combines a DVD recorder for compatible output and a HDD for flexibility. The HDD was added with no change in the form factor. One of the advantages of this approach is the ability to do 2X dubbing.

0803.2.7 Projector

Sanyo has developed a ultra-short throw distance projector. This uses a combination of a mirror and lens. This has the ability to project to 80” from a distance of only 80”. The resolution is XGA with a brightness of 2000lm.

0803.2.8 Microdisplay

TI has implemented a micodisplay for a HDTV 3D display using a digital micro-mirror (DMD). This creates two images which can be seen with shutter glasses. This is implemented as eyeware and claimed to be a low cost adder to HDTV.

0803.2.9 Flat Display

National Taiwan Normal University and SiPix have developed a smart card which integrates a flat electrophoretic display to create a one time password card. This proposed as a credit card where the individual is not present – online purchases. The display used is a Microcup EFD. This display is claimed to be superior to other EPD due to is high pressure tolerance. The card works by pressing a button multiple times to get a new password. The password has only a short lifetime.

0803.2.10 Networks

0803.2.10.1 WPAN

SK Telecom, Kookmin University and Pusan National University all in Korea developed a WPAN platform architecture of handsets. This architecture develops a WPAN based on using the cellular network. The handset is connected to other handsets, digital devices or an access points which are managed by a PAN server. The PAN handset platform provides abstract WPAN APIs to the developer. WPAN applications can be developed using both the cellular network APIs and the WPAN APIs. Examples cited include mobile game and voice terminal. The PAN platform is to be launched by SK Telecom in Korea.

0803.2.10.2 Cable

Samsung has developed a middleware architecture which integrates the cable television network with the IP network over cable. The basis for this is a Hybrid Digital Cable Receiver (HDCR) they have developed. The middleware in the HDCR is based on OCAP. The advantage of this approach is that it supports both RF and IP traffic on the same network with different communications protocols. It is proposed that the HDCR can be purchased at retail by the consumer. This would enable consistent services to be purchased independent of the cable operator. It is claimed that there will be faster time to market and more cost effective services.

0803.2.11 Home Server

ETRI, Korea, has implemented a Mini-MCU, Multi-point Control Unit, which can reside on a home server. This distributed mini-MCU allows for a four-way video conference to be held. This is based on SIP. There is a video conference module which supports MPEG-4 and four codecs. The camera used is USB based. Other video sources are supported as input to the codec. This approach has the advantage of improving the video quality for the same bandwidth as a centralized MCU.

ETRI, Korea, has developed a multimedia services framework for integrated services in the home. This helps define the requirements for the home media server. Such a server was implemented on a dual core system with the Linux OS. The services framework is based on the Open Home Network Framework (OHF) which provides 3 layers with a service API that supports communications, broadcast, and appliance control for the home network.

0803.2.12 Devices or Products

0803.2.12.1 Gadget Language

ETRI of Korea has developed the Intelligent Gadget Markup Language ( IGML). The intelligent gadget is a smart object which can process information and cooperate with other gadgets. They are usually deployed in context aware situations and capture environmental data, such as in health care situations. IMGL represents context for the gadget and describes event-condition-action (ECA) rules in XML for gadgets to do event based services. It is the intent that service developers will write rules in IGML for what the intelligent gadget will do. This research is part of research into a Personal Life Log.

0803.2.12.2 UPnP

Chungnam National University, Korea developed a bridge component which enables BANs (body area networks) and PANs (Personal Area Networks) to be discoverable with UPnP. The bridge is a wearable middleware framework using a NPnP – Network Plug and Play – which supports the UPnP discovery protocol. With this it is possible to bridge from Bluetooth to UPnP. The NPnP is only 57kB and is able to control Bluetooth devices in an actual environment.

ETRI, Korea and Sungkyunkwan University have implemented a protocol which allows for the control of home A/V equipment when that equipment is distributed across multiple homes and with a NAT between. This is an extension to UPnP with a Personal Mobil A/V Control Point (PMCP) and a Upnp relay. The implemented PMCP enables users to control UPnP A/V devices on their home and moving UPnP network at the same time, with zero configuration. This allow for the playing of media flies on a media servers on each UPnP network.

0803.2.13 Human Interface

0803.2.13.1 Gesture

A technique was developed which combines speech and gestures as a form of HCI. The system supports 10 natural gestures. A modified Time Delay Neural Network is proposed to support the fusion of audio and gesture. Experiments were conducted with 10 commands that used both speech and gestures. The fusion system showed a significant improvement is S/N ratio and an improvement over a ASR, automatic speech recognition, based system.

0803.2.13.2 Touch

ETRI, Korea, developed a virtual touch interface. This is based on augmented reality where the virtual touch screen is seen by the participant with a head mounted stereo display. The system detects and tracks the user’s hand position and perceives the finger tip. Select operations are accomplished by the alignment of the finger tip and the virtual display screen. The timing of the presence of the finger on the screen determines if it is a click or select operation. Mouse emulation is accomplished for most operations.

0803.2.14 Multi-View Video

Florida Atlantic University is proposing a HCI method to select video from multi-view video. This uses a motion sensing controller, like the Wii controller, to capture movement as a means of selecting the video view. A Wii controller was used to accomplish the motion detection and button functions in the demonstration system. Multiple control and navigation functions were accomplished with this interface.

Chinese Academy of Science and China University of Mining and Technology examined the HCI implications of user patterns of watching and selecting multi-view video. Head view tracking was used to determine viewing patterns. This gave information on view switching. From this it is possible to better design server systems for multi-view video.

0803.3 WAVE Comments

One had to only look at the audience and the presenters to realize that CE is in Asia with a few participants from Europe (dominated mostly by Philips). Our estimate that the distribution is in this order:

     Japan
     Korea
     Taiwan
     China

One could count on one hand the papers from the US.

This distribution certainly reflects the products seen on the shelves. The emergence of Taiwan and China indicates that these countries could begin to challenge the current leaders – Japan and Korea.

Overall we assessed the quality of the papers as high. Language was a barrier, especially in the question session.

With the exception of Apple there is no US company of any scale which participates in the CE market.

Return to the top

--------------------------------------

To subscribe to WAVE go to

http://www.wave-report.com

To unsubscribe also use the Wave Report Home page or send the preformatted UNSUBSCRIBE message:

List Management - Unsubscribe

Previous issues of WAVE, as well as other info can be found at

http://www.wave-report.com
http://www.3dlinks.com

Comments on or questions about the WAVE may be sent to:

Fourthwave Staff

or the below individuals below:

John N. Latta - Editor-In-Chief

Michael Robertson - Web Editor

The WAVE Report may be redistributed in full for individual readership and posted to newsgroups, Web, and FTP sites. This publication may not be reprinted or redistributed for profit. Short quotes are permitted but must be attributed to the WAVE Report. 4th Wave retains the copyright to the WAVE Report.