1. Home
  2. Electronics Design
  3. New Algorithms Increase Channel Bit Rates

New Algorithms Increase Channel Bit Rates

Algorithms that allow communication channel bit rates near the theoretical maximum at high signal-to-noise noise ratios (SNR) have been developed. Twenty five percent or greater increases are possible at SNRs typical in cable and hard disk drives.

C. Shannon, the founder of Information Theory, derived an equation that gives the absolute maximum bit rate that can be transmitted through a communication channel at any given SNR (known as Shannon channel capacity or SCC). Algorithms, called turbo codes have been developed that achieve SCC at low SNRs (e.g., 5 db). However, these codes do not achieve SCC at higher SNRs (e.g., 20 db).

Neural Systems Corporation (NSC) developed algorithms to process QPSK and hard-disk drive read channel outputs that allow near SCC rates for communication channels and near maximum possible disk-drive recording densities when the SNR is 18 db or greater. (QPSK and disk drive algorithms are virtually identical). For example, when the SNR is

27.5db the NSC algorithms are approximately 40% better than 256 QAM. Read channel model experiments have demonstrated SCC when Eb/No = 14.5db and the recording densities were 4.25.

For additional details see the NSC Web site at:

formatting link

Reply to
bsmithtech
Loading thread data ...

IMHO there's no free lunch.

If these algorithms work at only HIGH signal to noise ratios, that kinda seems to imply the DONT work at lower ratios.

So your disk drive will be 25% bigger, sure, but if the signal quality degrades, all of a sudden you'll start seeing bad blocks. Meanwhile disks using older algorithms will still be able to read the data.

Just like digital TV, the signal quality is nice until something gets a bit out of whack, then the quality goes to pot in a heck of a hurry. Many people would rather have a picture that degrades gracefully, instead of one that pixellates and blanks out all of a sudden.

Reply to
Ancient_Hacker

So why not just add noise? Was it needed a new algorithm?

Reply to
Mochuelo

On the website cited,

formatting link
, NSC claims to have developed logic that is trained like a neural network. In natural neural networks we find examples of extremely efficient neural coding that approach theoretical limits. The secret: continual adaptation to prevailing circumstances. In the future we can expect information-processing systems that reorganize themselves as the hardware ages, optimizing the bit-rate, system reliability and degrading only slightly until a final total failure, which would only lack grace if it hadn't taken so long to actually occur.

-- Joe Legris

Reply to
jalegris

...but the FCC has forbidden analog TV in the near future, so get used to trash..

Reply to
Robert Baer

Idiot! If one can sedn at a high SNR, during the travel to the receiver, and at the receiver, noise is accululated, lowering the SNR - making it harder to get a clean signal.

Reply to
Robert Baer

Actually, there are times and places where intentionally adding noise can dramatically improve restults. Especially in high precision D/A converters.

--
Many thanks,

Don Lancaster                          voice phone: (928)428-4073
Synergetics   3860 West First Street   Box 809 Thatcher, AZ 85552
rss: http://www.tinaja.com/whtnu.xml   email: don@tinaja.com

Please visit my GURU\'s LAIR web site at http://www.tinaja.com
Reply to
Don Lancaster

Why?

First, learn to write.

Second, learn to express yourself with clarity. The fact that both the channel and the receiver add noise to the signal is NOT conditioned to "if one can sedn at a high SNR."

Third, thank you for enlightening us with such an invaluable piece of knowledge, about how noise makes it harder to get a clean signal. Hadn't it been for you, I could have never guessed that.

And finally, it is obvious that you still don't understand why I wrote my first question.

Reply to
Mochuelo

Almost all audio CDs include noise (dither) that has been intentionally added when creating the commercial 16-bit copy from the

24-bit master.
Reply to
Mochuelo

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.