Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Qualcomm Incorporated (QCOM) -- Ignore unavailable to you. Want to Upgrade?

To: engineer who wrote (24387)	3/18/1999 3:48:00 PM
From: Sawtooth	Read Replies (2) \| Respond to of 152472

<<3/18 (Press Release) Today Ericsson displayed the T18 mobile phone. This dual band phone retains the classic Ericsson look and is packed full of new and interesting features, among them voice dialing, voice answering, active flip and vibrating alert. The phone is available in three colours: Granite Grey, Juniper Blue and Maple Red......>> Voice answering!!! What will they think of next!!! ; )

To: engineer who wrote (24387)	3/18/1999 5:10:00 PM
From: Ruffian	Respond to of 152472

Engine, Your Language> Coding, tools hamper 3G cellular telephony By Stephan Ohr EE Times (03/18/99, 3:11 p.m. EDT) PHOENIX — Better error-coding techniques will be needed for third-generation (3G) cellular telephony, and in fact, for any kind of communications service involving multimedia transmission. For cellular telephony in particular, error coding will help prevent garbled speech and images on the loss-prone airwaves. But current error-coding techniques — along with the tools for modeling them — are largely inadequate, UCLA professor John Villasenor said in his plenary talk at the International Conference on Acoustics, Speech and Signal Processing (ICASSP) on Tuesday (March 16). Villasenor called for an interdisciplinary approach to 3G cell phone design. The biggest problems communications systems designers have are examining trade-offs across Open Systems Interconnection layers, and "aligning a standard, a market and a product simultaneously," he said. "Too much effort is devoted to asking whether you can use the cell phone to transmit 128 kbits/second or 1 Mbit/second," he said, "but not enough attention is paid to what's happening out there — especially on the applications layer." Villasenor's own work with the standards-building International Telecommunication Union (ITU) concerns error-coding techniques that would allow a 3G cell phone to receive video or graphics-rich Internet material. To transmit video over a noisy, loss-prone medium such as a wireless cell phone, some sort of expansive error-coding would have to be inserted in the data stream. The code allows the receiver to check the integrity of the data stream, and perhaps reassemble it where significant errors are found. The amount of error coding required depends on the integrity of the cellular channel, Villasenor explained. The better the channel, the less coding is required. On a 3G cell phone transmitting video, the amount of error coding would be significantly greater than the actual data stream. For a bit-error rate of 10-2, for example, more than 75 percent of the transmitted bits would have to be devoted to error coding, Villasenor said. For a bit-error rate of 10-3, 85 percent of the transmitted bits would need to be devoted to error coding. Further, the Huffman coding techniques and other variable-length error-correction codes currently used for H.263 may not provide "the best tools" for correcting MPEG-4 video streams, Villasenor said. The Wideband Code Division Multiple Access (CDMA) that will likely be the standard for 3G generates very long and complex data packets, Villasenor said. Most error-coding techniques would drop one or more data packets upon the discovery of errors in the data sequence. With compressed video, that would produce dropped frames and visible splotches in the image. "Huffman coding produces severe error propagation when transmitted on a noisy channel," Villasenor said. The trick, Villasenor said, is to use "reversible variable-length codes." These are effectively symmetrical expansion codes that could be inserted both at the beginning and at the end of a data packet. Thus, the discovery of an error in the data sequence would not automatically require that the received data sequence be discarded. Rather, the reversible (or "look back") codes at the end of the data sequence would allow a partial reconstruction of the data — resulting in fewer dropped or garbled packets. Villasenor did not have definitive recommendations for specific expansion codes, but said that codes with exponential distributions, like those devised by Golomb-Rice, might work best. "The good thing about video," he said, "is that there is a high degree of redundancy." Parts of every frame or image could be dropped or reworked without noticeable impact on the received data stream. Thus, "hooks" for sophisticated error coding could easily be embedded in the video-compression algorithm. Villasenor said he is concerned, however, about the lack of availability of good modeling tools for DSP-based communications research. A major problem continues to be the modeling of effects from converting a theoretical floating-point algorithm into a practical fixed-point implementation. While Villasenor declined to criticize Cadence's SPW or Synopsys' Cossap, he said he found more model-building utility in the DSP Canvas tool from Angeles Design (San Jose, Calif.). While there are tools that support compression-algorithm development or channel coding or circuit design, there are no tools that support the analysis of communications systems across design domains, Villasenor said.