CD Shareware Magazine 1997 January

home *** CD-ROM | disk | FTP | other *** search

/ CD Shareware Magazine 1997 January / CD_shareware_1-97.iso / DOS / COMUNICA / V34PLUS.ZIP / V34PLUS.TXT < prev

Wrap

Text File | 1996-10-02 | 24.2 KB | 486 lines

==================================================================== V34PLUS.TXT ==================================================================== ANALOG ASYNCHRONOUS COMMUNICATION AT 33.6K BPS 30 SEP 96 By Paul Munoz-Colman FunStuff Software http://Ourworld.CompuServe.Com/Homepages/FunStuff_Software/ BACKGROUND. ~~~~~~~~~~ What are the terms V34, V.34+, and VFC? ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ V.34. The newest international asynchronous data communications standard for analog modems in today's marketplace is "V.34". V.34 was ratified in the Summer of 1994 by the international standards body ITU-T (formerly known as the CCITT). As originally published, V.34 offered modulation rates up to 28.8k bps (28,800 bits per second), with the opportunity for future extensibility. V.34+ That future is here today. Recent technology improvements enabled the implementation of optional features of V.34. These improvements, crafted by AT&T, US Robotics, and others, have been temporarily designated "V.34+" in today's marketplace (the ITU-T hasn't announced the final "V" label, but it is likely that the new improvements will still be called "V.34".) The improvements raise the maximum modulation rate from 28.8k to 33.6k bps. A typical 2,400 bps increase in connection rates occur, as well as improved performance at lower speeds. Even though they are still unratified, the 33.6k bps options which are now implemented in millions of modems, are well understood and well documented by major industry suppliers. As such, V.34+ is living proof of the high value of a well-designed, extensible, open systems communications standard. VFC For historical purposes, the early work on V.34, which began about four years ago, was called "VFC". "FC" stands for "Fast Class", and was an early, proprietary, and less robust implementation of V.34. What has happened to technology?! ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Older modems operated at 9,600 bps, using the V.32 protocol. The next generation of modems operated at 14,400 bps, and used the V.32-bis protocol. The V.34+ protocol operates at a top speed 2.5 times that of V.32-bis, and 3.5 times that of V.32. In sharp contrast to the extremely volatile marketplace that existed when I wrote the previous version of this article ("Asynchronous Communication at 28.8k bps, dated 17 Dec 94"), as I predicted then, the dust has settled. Nearly all vendors have V.34 modems on the market, and many have V.34+. There is still some high-speed chaos. Three years ago, prior to V.34, much haste occurred to get 28.8k bps modems to the marketplace. There were many different early versions of 28.8k bps, even within the same manufacturer of modems. This haste caused extraordinary confusion and lack of interoperability between modems. Connections were difficult to establish difficult to maintain, and unreliable to transfer data. Early implementations of V.34 also suffered compatibility problems. These problems were due to differences in the interpretation of the high complex specification. Some modems still suffer from this problem today. What is the big deal about V.34's implementation? ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ V.34's design limits. It is not only perfectly normal, but even typical in a V.34 connection to see less than 33.6k bps in a connection. As V.34 is not a fixed-speed standard, it makes or changes its connections based on phone line quality. It is very rare to get consistently perfect connections. Speeds of 33.6k bps require pristine phone line quality along the entire length of the connection. However, V.34+ is certainly capable of pushing the practical and ordinary limits of analog phone lines, commonly offering connection speeds of 24k, 26.4, and 28.8k bps. Analog modems communicate over voice-grade (unconditioned) telephone lines. To support speech, the minimum bandwidth (or "bandpass") of a voice-grade line must be at least 3,000 Hz (cycles per second). To make 3,000 Hz available on the phone line, the laws of physics require supplying more than 3,000 Hz of bandwidth. Because more than 3,000 Hz is available, the technological implementation of advanced mathematics in modems can make use of that extra bandwidth to give greater connection speeds. Where does this "extra" come from? To supply the 3,000 Hz in analog circuits, there must be a natural "rolloff" in the amplitude of the signal. This rolloff occurs at the lowest frequencies (near 0 Hz or DC), and at the highest frequencies (near 4,000 Hz). Even at the reduced levels, there is significantly usable available bandwidth, but under a variety of line conditions. The phone lines are pushed to the limit, by applying various mathematical techniques to compress the signal. This achieves the highest possible rates, at nearly theoretical limits. V.34 employs a smart method, called a "channel probe," which measures the frequency response and signal-to-noise ratio at various points across the bandpass. During the "handshake," modems send a series of tones to each other, at known signal levels and at defined frequencies. The modem calculates the level of the received signal at each frequency. By a set of rules defined by the V.34 protocol, the modem determines the maximum bandwidth available for use. So, just how good does a line have to be?! ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ To obtain and maintain a 33.6k bps connection, it takes a clear line which does not drop below about -44 dB (deciBels) or better, measured very close to 4,000 Hz, the upper limit of the rolled-off portion of the bandwidth. (-44 dB is the sound level of a clearly whispered conversation across a medium size room.) At -46 dB and below, modem receivers start to "go deaf." But at 4,000 Hz, the typical long distance telephone connection can be much quieter than -46 dB. At 4,000 Hz, it is not unusual to see rolloffs of -55 dB to -75 dB, which is closer to the background hiss level of a factory fresh medium grade audio tape. Standard transmit levels for domestic (US/Canada) modems are approximately 10 dB below reference level (-10 dB). During the initial transmission attempt, the actual transmit levels are negotiated. Receive levels vary widely, depending on the conditions of the local phone line, the line at the remote modem, and the long-distance or inter-office carrier facilities. Typical receive levels range from -40 dB at the low end, to -15 dB at the high end, with the -20 dB to -35 dB range being most common. Extreme values in either direction probably indicate a problem in the connection from your modem to your local phone company, which in some cases the phone company may be able to adjust. So how come V.34 is so robust? ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Recovery from adverse line conditions. The goal of 33.6k bps modem protocol is a simple one: under inevitably changing conditions, it should have a high top speed, and should spend as much time as possible operating at the highest possible speed. The V.34 protocol has advanced procedures for "training" (synchronization between modems), and for recovery from transient disturbances during training. There are several retrain and speed switching procedures to ensure the integrity of the link under adverse conditions. The line (channel) probe. V.34 "probes" the phone line for quality. The line (or channel) probe quickly examines line conditions and selects the best transmission strategy to optimize data transmission (there are a variety of such strategies available). This examination consists of measuring the amplitude of the signal at various frequency levels ("frequency response") and aspects related to signal distortion. This concept can be captured by instrumentation within a modem. In the case of US Robotics modems, visible instrumentation is built into the modem's command set, and the data from the commands is recallable to the user. Live data on the condition of a connection can be captured in numeric form. The numeric data can then be passed to Joe Frankiewicz's fabulous program called USRSTATS, which draws a graphical representation of the line's frequency response. USRSTATS is available on my web site, on Joe's BBS, and on USR's BBS or web site. The drawing below is the frequency response from a typical long distance connection. This one was from my home in Reston, Virginia, to the US Robotics bulletin board in Skokie, Illinois, using MCI as a long distance carrier. The rate for this particular connection was 24,000 bps (receive channel), and 19,200 bps (transmit channel). In this example, note the modem receiver "went deaf" above 3,000 Hz, so there wasn't much opportunity for highest speeds here. The measured signal level (in dB) is shown on the left Y-axis, and the corresponding attenuation (or drop, also in dB) from the practical usable signal level is shown on the right Y-axis. The frequencies in the bandpass (in Hz) are shown along the X-axis. ==================================================================== The V.34 Channel Probe ┌───────────────────────────────────────────────────────────────┐ │ -28 │ ∙ ▄ ▄ ▄ ▄ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ │ 0 │ │ -30 │ ∙ █ █ █ █ █ █ ▄ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ │ 2 │ │ -32 │ ∙ █ █ █ █ █ █ █ █ █ ▄ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ │ 4 │ │ -34 │ ∙ █ █ █ █ █ █ █ █ █ █ █ ▄ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ │ 6 │ │ -36 │ █ █ █ █ █ █ █ █ █ █ █ █ █ █ ▄ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ │ 8 │ │ -38 │ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ │ 10 │ │ -40 │ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ ▄ ∙ ∙ ∙ ∙ ∙ ∙ ∙ │ 12 │ │ -42 │ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ ▄ ∙ ∙ ∙ ∙ ∙ ∙ │ 14 │ │ -44 │=█=█=█=█=█=█=█=█=█=█=█=█=█=█=█=█=█=█=█=▄=∙=∙=∙=∙=∙=│ 16 │ │ -46 │ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ ∙ ∙ ∙ ∙ ∙ │ 18 │ │ -48 │ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ ∙ ∙ ∙ ∙ │ 20 │ │ -50 │ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ ∙ ∙ ∙ │ 22 │ │ -52 │ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ ∙ ∙ ∙ │ 24 │ │ -54 │ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ ▄ ∙ ∙ │ 26 │ │ -56 │ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ ∙ ∙ │ 28 │ │ -58 │ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ ∙ ∙ │ 30 │ │Level└───────────────────────────────────────────────────┘Atten│ │ dB 0 0 0 0 0 0 1 1 1 1 1 1 1 2 2 2 2 2 2 3 3 3 3 3 3 dB │ │ 1 3 4 6 7 9 0 2 3 5 6 8 9 1 2 4 5 7 8 0 1 3 4 6 7 │ │ Freq 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 Freq │ │ Hz 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Hz │ └───────────────────────────────────────────────────────────────┘ Frequency, Hertz (Hz) ==================================================================== The Channel Probe determines proper connection speed. V.34 measures signal levels at 25 frequencies across the entire channel, in intervals of 150 Hz. This provides a highly accurate sample of the channel bandwidth, and in selection of the appropriate "symbol rate." Because of this close spacing of the probe samples, the accurate profile (and its ability to provide problem detection) is a main reason why V.34 connections are so reliable. The channel probe occurs during initial modem negotiation, and during training and retraining. Additionally, the line's quality and noise levels are measured repeatedly during the connection. One of the objectives of the probe is to detect certain unusual non-linear distortion mechanisms present on some phone circuits, particularly international ones. The modems can then select the operational modes that better combat distortion. Using these kinds of measurement tools, with practice and perhaps some technical consultation, it is possible to become adept at determining different kinds of problems with phone lines. Even better, if you are lucky enough to have a cooperative local phone company or long distance carrier, the tools can even be used to help them troubleshoot and pinpoint adverse situations. However, please be aware that on unconditioned voice grade lines, Ma Bell and the long distance carriers are not required by law, statute, or tariff to "fix" this "problem." Why not? Because it is not really a "problem", but simply a fact of nature and of technology. Besides a quality line probe, V.34 does a cooperative (and nearly instantaneous) speed shift, also called a "fallback," which host computers can tolerate well. This rate renegotiation procedure allows rapid switching ranging from 4.8k bps up to 33.6k bps, as line conditions vary. Rate renegotiation is a tremendous improvement from earlier modem "retrains" (where the modems would isolate themselves from the host computer for up to a minute, while they recomputed the line parameters). Unlike renegotiations, host computers do not tolerate retrains well at all. Often, they think the line has gone dead, and so will themselves disconnect from the modem. So why does it get bad? ~~~~~~~~~~~~~~~~~~~~~~ Simple line impairment. Variations in line quality are typically the culprit for low connect rates. Line impairments result in several conditions: . link time-outs (the error control protocol does not receive a block of data within its expected timeframe), . link naks (the error control protocol requests retransmission of the data), . blers (block errors, or errors in received error control protocol or data blocks), and . retransmitted data blocks. Everyone occasionally gets "a bad line" and has to hang up and call again to get a better connection. However, if you find that you never or rarely connect at rates above 19.2k bps, you will want to investigate the line quality of your connections. There are significant differences in the reliability and stability of modems from different manufacturers, and in their available instrumentation. Those which are most reliable come from manufacturers who take the greatest care in implementation of the V.34 standard. In many cases, modems so produced will operate reliably at speeds that are slightly lower than others. Some modem manufacturers place greater store in reporting exceedingly aggressive initial connection rates. Doing this allows the modem to report a high connection rate, knowing that the switching protocol will quickly adjust the rate down to proper operating range. Other reasons why V.34 is a robust standard. V.34 has a number of features that make it the most reliable communications standard published to date: . precoding (which changes the transmitted signal to reduce the effects of noise multiplication in adaptive equalization, compensating for severe amplitude distortions); . powerful multidimensional trellis coding; . constellation shaping (which gives greater immunity to noise); and . nonlinear coding (which changes the transmitted signal to improve operation in the receiver, thus addressing the problem of distorted signal peaks due to nonlinear circuit elements). VFC, and earlier ITU-T standards at lower speeds, kept both the receive and transmit channels operating at the lowest of the two speeds (and their associated symbol rates). A channel impairment in either the transmit or the receive direction would drop both speeds to the single level tolerated by the impairment. In contrast, a major improvement in V.34 is first international standardization of independent receive and transmit channel speeds (and their associated "symbol rates"). This allows the receive and transmit channels of the modem to adjust independently and operate at different speeds, thus making maximum use of available bandwidth in the face of channel impairments. (Historically, independent transmit and receive channel speeds were first introduced many years ago by US Robotics in their proprietary High-Speed Transmission (HST) modulation, and were submitted and eventually incorporated in V.34 by the ITU-T.) V.34 has more robust Trellis Coding in use by the modem's receiver and transmitter. Trellis coding is a mathematical operation performed on the transmitted data that improves the system's noise immunity. The type of coding may vary significantly when connecting modems from different manufacturers. V.34 supports a 64 state 4 dimensional coding scheme for high noise immunity. All right, you convinced me! I just bought a V.34+ modem and am still having problems! What can I do to get a better connection? ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ . Try calling a different location. Line quality differs from region to region, and it may be a problem with the lines or modem at the other end of a particular call. . Try connecting with a local call. Sometimes the connections within a long distance call can cause impairments. (If this isolates the problem, you can try switching long distance companies.) . Try plugging the modem to a different phone line or wall jack. . Try eliminating all telephone extensions, phone line surge suppressers, line switches, utility monitoring devices connect to the phone line, and anything else on the line with the modem. . If you know someone else in your area with a high speed modem, ask what type of connections they make. Try making the connection from their location. If you encounter the same low connection rates, the problem may be resulting from impairments along the lines running to the local telephone company or within your home or office. Your telephone company or a private communications consultant may be able to help. . If you're a troubleshooter by nature, and happen to own a US Robotics modem, download the USRSTATS program from my web site, their BBS or web site, capture some statistics, and methodically study your own connections! Dropped Connections and Rate Switching in Early Protocols. Earlier protocols could only switch rates down to 14,400 bps. If you connected using one of these protocols, and the line quality dropped below that allowable for a 14,400 bps connection, the modems would disconnect. If this occurred frequently for a particular call, disabling the protocol was a way to make a connection when calling that modem again. As line conditions warranted, establishing a slower modulation would allow the modems to switch to lower bit rates. If the modems did not allow rate switching, the connection would likely drop. In those severe cases, locking the modem to a lower rate could complete the call. Dropped V.34 Connections and V.34 Rate Switching. Even though V.34 rarely drops connections, they will occur during a call, when there is a sharp decrease in line quality. In contrast to the early protocols, V.34 modems will switch down to rates as low as 4,800 bps to compensate for these changes. If the loss of quality is extremely severe, however, even V.34 will drop the connection. ==================================================================== Summary: ~~~~~~~ Technical phone line bandwidth requirements, and how bandwidth and symbol rates are determined for a connection: As already stated, connection rates are based upon the phone line's available bandwidth. Modems use the channel probe to test the phone lines before establishing a connection rate, and then select the highest "symbol rate" allowable. V.34 and VFC modulations allow adjusting the symbol rate to any of six possible values, to obtain the best match with the available bandwidth. Other protocols only allow a single, fixed value for the symbol rate, regardless of the bandwidth of the link. A "symbol" is a waveform transmitted by the modem. The waveform contains a certain number of encoded bits of data to move across the link. The receiving modem decodes this waveform, recovers the package of bits, and re-assembles it. Noise levels in the channel determine the number of bits encoded in each symbol. Lower noise levels allow a larger number of bits per symbol. The rate at which symbols are sent is limited by the bandwidth of the channel. Symbol rate is directly related to overall connection speed. A higher "symbol rate" generally allows greater data transfer speeds, but requires greater bandwidth. Once negotiation determines a symbol rate, it remains constant. To maintain low error rates, by considering both the changing characteristics and the levels of noise, the modem adjusts the bit rate dynamically. The chart below shows the approximate bandwidth requirements for each symbol rate. Thus, based on the connections you make, and/or by the quality of the diagnostics contained in the better brands of modems, you can determine the approximate bandwidth detected by the modem. For each symbol rate, a connection can be made from the choice of frequency ranges. Thus, the modem selects the best quality for each call. These are maximum bit rates. V.34 will connect at speeds as low as 4,800 bps with any of these symbol rates. VFC will only connect down to 14,400 bps. If the bit rate is much lower than the maximum bit rate supported by the symbol rate, the phone line has lots of noise or other impairments on it. ==================================================================== Maximum Maximum Symbol Protocol Carrier Bandwidth Bit Rate Bit Rate Rate Range Frequency Requirements (V.34/VFC) (V.34+) ----- --------- --------- ------------ -------- -------- 2,400 V.34/V.34+ 1,600 Hz 400-2,800 Hz 21,600 21,600 VFC /V.34+ 1,800 Hz 600-3,000 Hz 21,600 21,600 2,743 V.34/V.34+ 1,646 Hz 274-3,018 Hz 24,000 26,400 VFC /V.34+ 1,829 Hz 457-3,200 Hz 24,000 26,400 2,800 V.34/V.34+ 1,680 Hz 280-3,080 Hz 24,000 26,400 VFC /V.34+ 1,867 Hz 467-3,267 Hz 24,000 26,400 3,000 V.34/V.34+ 1,800 Hz 300-3,300 Hz 26,400 28,800 V.34/V.34+ 2,000 Hz 500-3,500 Hz 26,400 28,800 VFC 1,875 Hz 375-3,376 Hz 26,400 3,200 V.34/V.34+ 1,829 Hz 229-3,429 Hz 28,800 31,200 VFC /V.34 1,920 Hz 320-3,520 Hz 28,800 3,429 VFC /V.34 1,959 Hz 244-3,674 Hz 28,800 V.34+ 1,959 Hz 244-3,674 Hz 33,600 ==================================================================== Permission is granted to reprint and redistribute this information only in its entirety. Greatful acknowledgement for selected source materials is given to: Joe Frankiewicz, Paul Gebert, and Dale Walsh of US Robotics, Inc. ====================================================================