Network Computingwww.networkcomputing.com

Get More Mileage From V.34 Modems

Most major modem manufacturers now offer V.34 modems. While the new standard promises data rates of 28.8 Kbps -- a doubling of the 14.4 Kbps provided by the earlier V.32bis standard -- there are significant differences in implementations by modem vendors. Prospective buyers need to exercise caution when choosing V.34 modems because the external features that typically make for a good modem, such as front panel buttons or bundled communications software, are not the best guideline. Instead, look closely at vendors' implementation of the V.34 standard.

V.34: A Standard Not a Seal

For this reason, there may be inconsistencies and variations in performance between one modem and another and from one phone call to the next.

V.34 is backward compatible with earlier modem standards. Officially, all V.34 modems are compatible with V.32bis, V.32 and V.22bis modems. In addition, a number of manufacturers' modems are also compatible with V.FC and V.32terbo modems.

V.34 supports a wider range of data rates. The V.32bis spec only provided data rate compatibility from 7.2 Kbps to 14.4 Kbps. On the other han d, V.34 supports data rates from 2.4 Kbps to 28.8 Kbps. This may not seem like a big difference but it's helpful to have a modem that can potentially back down as needed.

Under the hood of V.34 modems you'll find that fuel-injection technology like adaptive pre-emphasis, line probing, multidimensional trellis-encoding, warping and more are all part of the specification. These hot-rod features -- designed to squeeze out the maximum from the PSTN -- are not only complex but also leave options for each modem vendor to resolve. For example, the determination of what carrier frequency to use allows developers to choose from 11 possible combinations and six different symbol rates, each with two possible carrier frequencies.

So, how do users know which vendors are implementing the standard best? First, go with experience. Pick the vendor who has been at it the longest. Second, look for real-world test results. To this end, there is some help. The TIA has published guidelines for testing V.34 modems. The TSB-37A is a model of all of the connections possible in the continental United States. The TSB-38 specifies how to test given the TSB-37A model. This is designed primarily to help modem vendors but it can also be helpful in determining which modems to purchase.

Expected Performance

The connection speed doesn't really tell the whole story because a faster connection speed over severely impaired lines might result in a higher error rate, requiring more retransmissions and lowering overall thr oughput. The overall average throughput for all of the modems in the real-world test was 5.6K characters per second (cps). The best modems averaged 7K cps, with individual transfers occasionally peaking in the 9K to 11K cps range. The worst turned in average rates in the 2K cps to 4K cps range. In tests where the phone lines represented the best case possible, the average throughput was slightly higher at 8K cps for a single directional transfer and 6K cps for a bidirectional transfer.

The TSB-38 guidelines specify that if a modem experiences a high rate of errors due to an excessively aggressive connection speed-resulting in a high number of retransmissions with little or no characters successfully transmitted-the connection will be considered a failure and a result of "no characters transmitted" will be recorded for the test. This was the factor for modems that turned in the lower throughput rates. While individual tests may have been fast, some of the impaired line conditions caused the modems to fail altogether, lowering their throughput score.

Beyond Throughput

This testing indicates that there were significant differences between V.34 modems that became apparent only when subjected to real-world conditions. We were able to create the real-world conditions by using test equipment from Telecom Analysis Systems of Eatontown, N.J., the leading supplier of modem test equipment to V.34 modem vendors and accepted as providing results in compliance with V.34 standards. Prior to purchasing V.34 modems, you must consider real-world test results to determine how well a particular modem implements V.34.

External Features That Ma tter

For desktop or personal use, important features include software, cables and even a high-speed Universal Asynchronous Receiver/Transmitter (UART) interface board.

However, if the modem will be used for an industrial-strength, centralized dial-in/dial-out application, support of a modem-management application and the ability to transmit synchronous as well as asynchronous traffic is more important.

The inclusion of front panel buttons and menu displays is useful in situations where the Data Terminal Equipment (DTE) does not support an automated or scripted initialization of the parameters with which it will communicate with the modem. For example, a device that does not have a way to store a phone number must rely on the modem to dial a stored number. In this example, the use of a front panel can facilitate the storage and dialing of the required phone number. However, this is becoming less the case as all V.34 modems support AT command sets and all PC software supports the scripting of these initialization commands.

Maturity of Product Line

While there are internal differences among products and some instability issues to resolve, V.34 modems are clearly products worth considering. You're heading in the right direction if you research before buying and recognize that a pro-mise of V.34 on the box is not necessarily a seal that guarantee s V.34 performance.

Bruce Boardman can be reached on the Internet at bboardman@nwc.com.