TITLE : V.34 MODEMS: OFF TO A FAST START? AUTHOR : By Ken Krechmer DATE : April 1994 TYPE : Technical ( Extracted from Data Communications Mag. Typed by Rollerboy.) V.34 MODEMS: OFF TO A FAST START? --------------------------------- WITH COMPRESSION, V.34 MODEMS CAN TOP OUT AT 100KBIT/S, BUT INTEROPERABILITY ISSUES AND OTHER CONCERNS COULD SLOW DEPLOYMENT. At 28.8 kbits/s, the V.34 standard now under consideration by the ITU-TSS will double the data rate of the quickest standards-based modems now on the market - and approach the upper limits of the top transfer speeds possible on today's phone lines. And with data compression factored in, V.34 modems can achieve an effective throughput of 100kbits/s. There's more: Even before it has been completed by the ITU-TSS (International Telecommunications Union - Telecommunications Standards Sector), V.34 can be characterized by its unprecedented price/performance ratio. In April 1992, Motorola Codex offered a modem - a precursor to V.Fast - that could sustain a maximum data rate of 24kbits/s and sold for $1,395, which translates into $0.058 bit/second (1,395/24,000). At that time, 14.4kbits/s V.32bis modems were selling for about $1,000, which works out to $0.069 per bit/second. So the Motorola Codex product was priced very aggressively. Now Hayes Microcomputer Products Inc. and Microcom Inc. have announced V.Fast Class products at roughly $600 that can operate to 28.8kbit/s. This translates into $0.02 per bit/second - which should make the choice of true V.34 modems (when they arrive) an easy one to make. And bear in mind that faster modems also mean shorter calls and less time spent per call. Clearly, the new standard promises many rewards, but it won't be easy sailing for network managers, at least at first. When previous high-speed modem standards were introduced, it proved difficult to ensure that the earliest products were interoperable, and the complexity of V.34 - as well as its suite of optional modulation techniques - will make interoperability even harder to attain. With previous modem designs, such problems were not resolved until six to 12 months after production shipments commenced, but in this case it will probably take slightly longer. Moreover, there is no consensus among vendors as to the best alternative to PC and Macintosh serial ports, which are too slow for V.34. Software that redirects data to the parallel ports, or enhanced serial ports, seems like the early favourite. Finally, it remains to be seen whether such pre-V.34 modems as proprietary V.fast modems (including what vendors call V.fast Class devices) will still be in use after V.34 is finished. If they are, they will confuse the marketplace and hamper users' efforts to build interoperable networks, since compatibility among all products at higher data rates may not be possible. V.34 standards are due to be finished in June 1994; a technical complete recommendation should be available several months before then. Pre-V.34 products that can be upgraded to V.34 are already on the market, and true V.34 implementations will probably be introduced shortly after the standard is final. V.34 DEFINED Since 1991, ITU-TSS Study Group 14 - formerly SG XVII - has been developing a recommendation for duplex modems operating at rates above 14.4kbit/s, already supported by the V.32bis standard. This is the substance of V.34. In the U.S., the work is supported through the Telecommunications Industries Association (VIA) TR-30.1 committee. At present, the V.34 draft recommendation defines data rates up to 28.8kbits/s; the final draft may include data rates as high as 32kbits/s, which is technically feasible. Vendors, however, may think themselves better served by leaving higher data rates for a future modem standard or for proprietary products that will help them gain an edge on competitors. A decision was expected at a TR-30.1 meeting scheduled for the middle of December. To promote the best possible performance on phone lines, draft V.34 employs a variety of techniques that have not been used in current modems recommendations. One concerns start-up, when V.34 modems choose the best carrier frequency to carry data. In the past, modems could use only one frequency, even if there was interference on it. Unlike older devices, whose throughput was reduced by poor telephone lines, V.34 modems will be able to achieve maximum throughput by adapting themselves to the characteristics of the phone line. V.34 uses a variety of other new techniques to attain its high throughput. These include adaptive precoding and four dimensional coding. MODEM HANDSHAKING When one modem dials another, they go through a process called handshaking to determinate which modulation scheme to use. The method that today high-speed modems employ is defined in Annex A of the V.32bis standard, outlining the procedures for automatic compatibility among V.22, V.22bis, and V.32 modems. V.32 modems and V.32bis have the same negotiation process, so their automatic interoperability is already defined. Annex A handshaking procedures are based on the tones characteristic of each modulation start-up sequence. Detecting tones is time-consuming, and the more modulation techniques are supported the more time it takes. With today's high speed modems, which support some six modulation techniques, including V.32, V.22bis, and Bell 212, it occasionally takes more than eight seconds to establish a carrier. Complete handshaking, which involves settling on common error-correction and compression techniques, can take 20 to 30 seconds - about as much delay as users will tolerate. Because of the time it takes to detect tones, today's handshaking procedures are not appropriate for V.34 modems, which will support more than six modulation techniques. The plan is that such modems will work with the entire range of duplex dial-up modems, function with existing Group 3 facsimile modems, serve as the basis for new high-data-rate facsimile transmissions, disable network equipment that compresses the 64kbits/s PCM (pulse-code modulation) channels, and ultimately support videophone and other new services. V.34 thus requires a handshaking scheme different from those used with existing high-speed modems. It will include parts of another recommendation now being developed: V.8 - previously V.id - which is the best way any two ITU-TSS modems handshake in the future. Before the V.34 negotiation takes place, the V.8 negotiation, using a modulated calling tone and answer tone, transfers information about the two modems' functional capabilities. First, the Calling Menu (CM), a data sequence using V.21 low-band modulation (a handshaking technique originally developed for 300bits/s modems), is to be sent from the originating V.34 modem to the answering V.34 modem. This describes the range of functional capabilities the originating modem supports. The answering V.34 modem responds to the CM with a Joint Menu, or JM (using V.21 high-band modulation), indicating the common capabilities of the modem at each end. Once this is done, a probing signal is passed between the modems to identify impairments in the telephone channel. After receiving the results of the probing signal, the modem receivers and transmitters (in each direction) will start with the lowest speed and move up until reaching the highest speed at which they can exchange data. The goal is to permit V.34 modems to complete the entire start-up sequence and commence data transmission within five seconds. V.8 will not prevent a modem from working with current modems that use tone detection for handshaking. Although no vendor has announced plans to use V.8 with anything but V.34 modems, it can also be used in V.32bis modems (or any other modem) to reduce the amount of time it takes to handshake. DIGITAL INTERFACE ISSUES The high data rates proposed for V.34 will prove a challenge in other areas as well. For example, the EIA-232 serial interface widely found on PCs and Macintosh computers is specified to operate only up to 20kbits/s - considerably slower than V.34. That isn't a great problem however, because many implementations use lower voltage swings to support faster speeds reliably. (To represent data, EIS-232 is specified to use voltage swings from -15v to +15v. Lower voltage enables faster data rates; the trade off is that cables can't be as long as they might otherwise be.) The real problem is that PCs and Macs use the EIA-232 serial interface with a combination of UARTs (universal asynchronous receiver/transmitters) and character oriented communications software packages that are not reliable at data rates above 20kbits/s. Most computers don't have 16550 UARTs that provide buffers if the system processor is too busy to transfer data from the serial port; instead, they use older UARTs. Even if a computer does have 16550 UARTs, its communications software may not be designed to use the buffers or to provide constant service to the UART during multitasking. For V.34 modems to be employed reliably, new interfaces will be required. There are at least five possible solutions. The first is EIA-530A, an interface formulated several years ago to replace both EIA-232 and V.35. EIA-530A is backward compatible with EIA-232, so existing modems and cables can work with it, and it supports data rates to 2.1Mbits/s. Unfortunately, 530A has not received much attention from the PC industry, and it is doubtful that PC manufacturers, already hampered by slim profit margins, will choose this solution. EIS-530A will probably find some use with V.34 modems, but for such communications equipment as FEPs (front-end processors), T1 multiplexers, and routers. The second possibility, V.10, is an electrical standard that can use the same hardware as EIS-232 but has different electrical characteristics. V.10, which supports data rates up to 100Kbits/s, is electrically compatible with EIA-232. The problem is that while V.10 could be implemented in new serial interfaces, the UARTs and software would remain unchanged. Operation above 20Kbits/s would thus still be problematical. The third approach calls for V.34 modems to support the IEEE 1284 standard for bidirectional parallel ports - a desirable solution, because they are already installed in most PCs. This approach, however, is dependent on the willingness of manufacturers to provide software to redirect data intended for the serial port to the parallel port. Xircom Inc. (Woodland Hills, Calif.) has announced such software for Windows and DOS, while Microcom Inc. (Norwood, Mass.) has introduced a Windows redirector. A fourth possibility could be connecting via the PCMCIA (Personal Computer Memory Card International Association) interface, now available on most new laptops (though not on desktop PCs). This solution will have to wait until it is possible to implement V.34 on the PCMCIA form factor. Finally, a proprietary, enhanced serial interface - provided through an add-on card, with buffers and software drivers - could establish a direct connection to the PC AT bus. This solution is currently provided by Hayes Microcomputer Products Inc. (Atlanta), Motorola Codex (Mansfield, Mass.), and others at a cost ranging from $75 to several hundred dollars. WHAT'S IN A NAME? Two different types of proprietary high-speed modems may be confused with their V.34 counterparts. The first are proprietary V.fast implementations that offer higher data rates than V.32bis does and will be upgradable to V.34 after it is complete, when they will cease to exist. The second consists of extension to V.32bis in the form of V.32terbo, which provides for speeds above 14.4Kbits/s; these modems will survive as low-cost alternatives to V.34 after it is complete. Motorola Codex has announced a proprietary V.fast modem; Penril Datacomm Networks Inc. (Gaithersburg, Md.), Racal Datacom Inc. (Sunrise, Fla.) and General Datacomm Inc. (Middelbury, Conn.) have announced competing proprietary high-data-rate dial-up modem implementations. All four vendors say that their products will be upgradable to V.34. During fall 1993, a division of Rockwell International Corp. (Downers Grove, Ill.) announced V.fast Class (V.FC), a chip set implementation of its proprietary V.fast version, operating at data rates up to 28.8Kbits/s. This implementation, supported by a number of manufacturers, including Hayes and Microcom, does not use the V.8 handshaking mechanism, so without software modification, it cannot be compatible with future V.34 modems. One of the principal applications of V.34 technology is cutting the transmission times of Group 3 facsimile by providing for higher data rates. The highest data rate now supported in the Group 3 recommendations is 14.4Kbits/s utilizing the V.17 recommendation. Most facsimile machines support 9.6Kbits/s, based on the V.29 leased-line modem standard. Study Group 8, which approves Group 3 facsimile recommendations, is considering the idea of implementing a half-duplex V.34 to reduce the complexity and thus the cost of facsimile modems. It is also possible that a lower-data-rate full-duplex modem - possibly V.22 (1.2Kbits/s) or V.22bis (2.4Kbits/s) - could be used for the lower-data-rate negotiation phases of a facsimile connection. Currently, the low-data-rate negotiation phase (T.30) is accomplished using V.21 (300bit/s) half duplex. Current SG 8 schedules suggest that extensions to Group 3 to include V.34 could be introduced before the end of 1994.