Toshiba MPEG-4 video cell phone........
eetimes.com
ISSCC: Toshiba to describe video cellphone
By Anthony Cataldo
EE Times
(02/07/00, 11:13 a.m. EST)
KAWASAKI, Japan — Aiming to seed a new class of video cellular phones, Toshiba Corp. this week will detail an ultralow-power MPEG-4 processor targeting third-generation cellular handsets that could ship in Japan early next year. Toshiba has recruited CAD supplier Zuken Inc. to help deliver, as early as this spring, other key components for wireless visual telephony systems to ride the wideband-CDMA nets that Japan will turn on in March 2001.
Toshiba will describe its second-generation MPEG-4 decoder at the International Solid-State Circuits Conference (ISSCC) in San Francisco this week. Engineers will detail a 60-MHz chip that can decode QCIF-resolution video at 15 frames/second while consuming as little as 240 milliwatts.
Toshiba's move could be the first of several efforts aiming to showcase the capabilities of 3G cellular, which Japan is expected to be the first country to try out. But observers quickly noted that the pursuit of the video phone — even in wired forms — has left a field littered with failures. Even W-CDMA systems may strain under the bandwidth, power and cost requirements.
"Toshiba and Zuken will provide a solution with which engineers can easily develop TV phones based on conventional cell phone technology by just adding a camera and a liquid-crystal display," said Shigenobu Minami, senior manager of Toshiba's mobile computing and communications development center.
Toshiba has been working on its own MPEG-4 decoder device for more than three years. The company announced a 64-mW MPEG-4 core at ISSCC in 1998, developed an MPEG-4 video codec LSI last autumn and is to present the second-generation chip with both video and audio capability this week. The first-generation device is the one going into the Zuken evaluation kit that will roll out in March.
Realizing the power-budget constraints on cell phone manufacturers, Toshiba is using every architectural and transistor design trick in the book to wring out the wattage. These include a more efficient way to decode video, the smallest manufacturing process design rules available, special low-voltage logic transistors, handcrafted microprocessor cores and on-chip DRAM.
Integration story
Based on 0.25-micron CMOS technology, the 60-MHz chip can perform all the needed functions for encoding and decoding QCIF-resolution MPEG-4 video at 15 frames/s, along with G.729 speech and the H.223 multiplex protocol at 64 kbits/s. Power consumption is 22 percent lower than in a conventional design with a separate video codec chip, DSP speech codec and external DRAM, Toshiba said — down to 240 mW.
The company believes the device, scheduled to begin sampling this summer, should be sufficient for cellular handset makers to build wideband code-division multiple-access prototypes with an LCD and low-power CMOS imager.
Making a shrink
Early next year, Toshiba will debut the next version, an architecturally equivalent design based on 0.18-micron design rules. Moving to the new process rules should halve the die size from the current 117.5 mm2 and reduce power dissipation to less than 100 mW, the company said. It is this device that is intended to transform the volume-production cellular phone into a fully self-contained video phone.
"I'm sure we can do a video phone by next spring if there is demand," said a confident Tohru Furuyama, a veteran DRAM researcher at Toshiba and senior manager of the company's digital multimedia LSI development group.
Not all service providers and cell phone manufacturers around the globe may share his enthusiasm for introducing video services so early in the 3G game. But in Japan, where service providers have already seen high demand for e-mail and limited Web-browsing capability, video will be pursued with a vengeance, Furuyama predicted.
"The situation is different in Japan," he said. "Video or picture-oriented services are in more advanced stages [here] than in the United States or Europe."
Helping turn the chip into a ready-to-use cellular design is EDA partner Zuken, which vows to release an evaluation board kit for visual telephony in March. The company has already started taking orders for the kit, and promises to follow it up with a full-blown development kit. Zuken said it will also license the intellectual property (IP) used in the evaluation kit.
Besides the first-generation MPEG-4 video codec supplied by Toshiba, the evaluation board kit includes an audio codec DSP, FPGA implementations of multimedia multiplex circuitry and an LCD controller, and middleware to control communications. The kit will comply with the IMP-2000 3G-H.324M standard, Zuken said.
Though the kit provides a separate audio codec, "Customers should have their own audio codec that they use in their conventional cellular phone products," said Minami. The two companies intend to offer an audio codec as IP to those who want it. In the future, they will provide an evaluation kit equipped with the newest MPEG-4 chip with audio/video codec.
Showing off
Service providers and phone manufacturers, for their part, have been teasing customers at press conferences and trade shows with mock-up phones gussied up with a camera and LCD. But in the real world, there is no way for a teenage girl from Tokyo to pull a cell phone out of her purse and show a friend in Osaka the platform-heel boots she has just bought.
How to deliver on the long-promised video phone is a question those in the chip industry are now pondering as they work with their manufacturing partners to juggle a range of power and chassis design trade-offs. Build in a camera and LCD, and you add more processing power and require a larger battery — two no-no's in a market that places a premium on sleek, ultralight phones.
One obvious solution would be to add a beefier digital signal processor, the main engine in the digital baseband, and program it to do the decoding and multiplexing of MPEG-4, a lighter version of MPEG-2 designated as the de facto standard for wireless video transmission. That may happen eventually, but probably not for early versions of W-CDMA phones with video capability.
"There are some issues still to be resolved," said Panos Papamichalis, director of Texas Instruments Inc.'s Tsukuba R&D Center in Japan, which at this moment has more researchers working on wideband CDMA than on any other application. "Initially, I would like to see people have some improved quality of voice transmitted."
For new applications like video, Papamichalis said, "there's going to have to be some hardwired functions for the first generation. When you have a high-Mips device, it consumes too much power, so the first generation will probably see some more hardwired portions. This usually reduces flexibility."
Questionable demand
Herschel Shosteck, president of market research firm Herschel Shosteck Associates Ltd. (Wheaton, Md.), said the wireless video phone is more of an economic issue than a technological one. "In order to deliver error-free wireless video service, the service providers will have to increase their cell density. At a minimum, they will have to double the number of cell sites, and more likely they will have to triple or quadruple them," he said. "That is very, very expensive, and that's why you won't see this happening any time soon."
Even more important, Shosteck doubts whether there will be a market for the technology. "The capability has been around for some time, but not many people are using it," he said. He pointed to a variety of social and cultural barriers, including the simple fact that many people don't always want to be seen by the people they are speaking with. "If nobody is using video phones on a land line, why would they want to do it on a mobile phone?"
However, Doug Bailey, director of marketing for embedded products at 8x8 Inc. (Santa Clara, Calif.), predicted that other applications will drive the market for wireless video telephony. One growing field is security cameras, where non-real-time video traffic is in demand and won't tax existing cellular capacity. Other uses include emergency medicine — people who need to see what is happening and would benefit from mobile video capabilities.
"There is no doubt in my mind that wireless video services will be a huge market, but it won't be driven by people making video phone calls," Bailey said. "There won't be much demand for Dick Tracy-style video wrist communicators."
One remaining issue is the telephony protocol that video traffic will use. Bailey said that most existing wireless packetized video is already using the H.323 standard. Linking a handset to a land line means converting the MPEG-4 packets to the H.323 format, which he said can add as much as 120 milliseconds to the call delay.
— Additional reporting by Yoshiko Hara and Will Wade |
