Floyd Davidson <floyd@ims.alaska.edu> (06/20/91)
In article <telecom11.464.1@eecs.nwu.edu> think!barmar@bloom-beacon. mit.edu (Barry Margolin) writes: > In article <telecom11.463.2@eecs.nwu.edu> miller@dg-rtp.dg.com (Mark > Miller) writes: >> OK, this one has been bugging me for some time, but what is the highly >> touted "Fiber Optic Quality"?? >> Now as I understand it, virtually all long distance calls are sent in >> a digital format. Last time I checked, digital data didn't really >> care whether it was sent by copper, microwave or fiber. > Of course it does. Digital data can be corrupted by static or > crosstalk on the line, just as analog data can. Computer > communication generally uses error detection and retransmission on top > of the digital medium in order to produce error-free transmission. > Audio telephone communication is generally more concerned with > transmission speed than fidelity, because listeners are able to deal > with minor errors. Higher fidelity lines mean fewer such errors. The measure of quality on a digital system is the bit error rate. There is no significant difference between the bit error rate of fiber systems compared to other digital systems. In all cases it is specified at something like 10e-5 and typically runs at one or two magnitudes better than that. The usual "turn-up test" done when installing digital systems is a 24 hour run with a zero bit error count. "Fidelity" on the digital part of a digital system has no meaning. The "normal" types of degradation found on digital systems are frame clock slips and high bit error rates. In the case of a single bit error it is not going to be perceptable to the human ear. It is also not going to cause a hit on a modem connection using that line. Framing slips are caused when the transmit data is being sent faster or slower than the receive system is accepting the data. It can either result in one byte being lost or one byte being repeated, and will not normally be detected by a human ear. It will not normally cause data hits for modems operating at 1200 bps or less either, but it does cause a phase jump, which will be detected by 2400 bps or higher modems. (If one byte is lost it is 1/8000 of a second of sound that is different than what it should have been. It may or may not be much different, either.) There are some digital systems that do provide error correction. Most satellite modems use forward convolutional error correction. However this does not necessarily mean the output is exactly the same as the input, but rather that the error rate is going to be less than that 10e-5 figure. The problems one is likely to find with using digital systems, whether fiber or other, are related to data compression. Packet trasmission systems can compress four or five to one, and there are many circuits being provisioned at 32 kbps rather than the 56 kbps or 64 kbps that we tend to think of as one voice channel. High speed modems work well over 56 kbps, and not too well with the lower bit rates or with compression on packet systems. The highly touted "fiber quality" is a marketing ploy. In fact what fiber gives is quantity and cost effectiveness compared to other digital transmission systems. All of them give the same increase in quality compared to analog systems. Floyd