September 1998

The Future Of Fax

BY BROUGH TURNER

There are over 100 million fax devices in the world today. More than half the "telephone" traffic between North America and Asia is fax, not voice. And the number of fax machines sold continues to grow 25 to 30 percent each year. There is no doubt that fax is a significant form of communications!

But today’s fax is inefficient. Therefore, it is more expensive than necessary. Sending a fax ties up 128 Kbps of bandwidth (64 Kbps each way) just to do some negotiation, at 300 bps, and then transmit some relatively low-resolution images, one way, at (typically) 9600 bps. An e-mail attachment sent over the Internet is more flexible and far more efficient. So, will the phenomenal growth of the Internet, e-mail, and e-mail attachments make fax machines obsolete? Hardly.

There will be changes on the fax landscape, but fax machines will remain. Like telephones, fax machines are simple to use. You don’t need a computer (and, more important, you don’t have to learn how to use a computer) to send a fax. A fax machine is the essence of simplicity. You put paper on the machine and dial. Whatever happens to the fax machine, this simplicity must remain. But how will fax evolve?

If you were able to start from first principles today and design a fax service (fax machine and fax standards), you would use the Internet. The "fax machine" would plug directly into the Internet; that is, it would connect to 10Base-T Ethernet on corporate LANs, or to whatever home equivalent emerges. The Internet fax (I-fax) machine would still be simple to use — just put the paper down and dial. But there are 100 million fax machines in the world. Current fax standards are already well established. We can’t design I-fax from scratch. We have to start with existing technology.

A SIMPLE BEGINNING
Fax required many decades to become standardized, cost-effective, and ubiquitous (in that order). Fax is actually older than the telephone. The first fax patent was granted in 1843. By the 1930s, systems with rotating drums were widely used to send photographs between news bureaus. Still, there were no standards, and the technology was slow and expensive.

Standards Progress
The first standards appeared in 1966. However, the major manufacturers continued to use their own modulation schemes. In 1978, the predecessor of today’s International Telecommunica-tions Union (ITU) introduced Group 2 fax standards. These were adopted by most manufacturers and helped start the widespread deployment of fax technology.

The real excitement happened in the 1980s. The Group 3 digital fax standard appeared in 1980. It worked over ordinary telephone lines and provided adequate quality for most business purposes. Competition among modem vendors caused the price of modems (an essential component of every fax machine) to drop. Reasonably priced desktop fax machines emerged. As more and more people bought fax machines, the utility of fax service increased, following Metcalfe’s law of networking: connect any number, n, of machines — whether computers, phones, or fax machines — and you get n2 potential value. Indeed, during the 1980s, the value of fax grew far more rapidly than simple subscriber counts would suggest. By the end of the 1980s, fax was a ubiquitous business tool.

Computer-Based Fax
The advent of low-cost PCs led to computer-based fax, in workstations and in servers. Image quality improves when a computer directly converts a document to fax image format, avoiding alignment, round-off, and other errors of the scanning process. And sophisticated software in LAN fax servers provides simplified administration, distribution lists, labor savings, and telephone cost savings.

But G3 fax has changed little since 1980. Modems have become faster and fax machine costs have come down. And, most important, each successive improvement in fax technology has been backwards-compatible.

THE FACES OF CHANGE
Today, the Internet is about to change the way fax works. But two things are certain. The simplicity of the fax machine will be preserved, and the new system will interoperate with the installed base.

For now, Internet fax deployments are being driven by cost savings. I-fax is much more efficient with bandwidth and resources. In the future, we will see that I-fax is also more flexible and easier to integrate with other applications.

But using the Internet brings up all kinds of issues. How do you "just dial" an Internet address, for example? Will the I-fax machine include a full alphanumeric keypad or a simple keypad with a complex method for entering speed dial numbers? How does an I-fax machine connect to a traditional fax machine? And how does an existing machine connect to the new I-fax machine?

New IP Fax Machines
Products — both I-fax machines and I-fax gateways — are already emerging, pre-standards. I-fax standards discussions are well advanced within the IETF and the TIA, with their results feeding into the ITU. The first set of emerging standards are called T.37 and T.38.

T.37 Development: T.37 uses the TIFF-F image format for fax, but uses e-mail over IP networks as the transport mechanism. The T.37 group settled on the TIFF-F image format because it was already defined and in use. More contentious is the method for confirming delivery. With traditional fax service, you know in real-time whether your transmission was successful (typically, while you are still standing by the machine). By contrast, today’s e-mail systems don’t provide delivery confirmation beyond the local domain, or at most, beyond the last mail handler (for example, a POP3 server). So the I-fax standards groups must agree on new techniques (or unfortunate compromises). The fast-track T.37 development anticipates interoperability testing before the end of 1998. Expect to see T.37 PC-based I-fax solutions and I-fax machines this year.

T.38 Development: T.38 defines real-time fax. Real-time fax uses the existing T.30 fax protocol, so I-fax can interoperate with traditional fax machines. Unfortunately T.30 was defined at a time when memory was expensive, so T.30 avoids storing images. As a result, sending and receiving G3 machines must run in a lock-step fashion. The lock-step approach, however, is difficult to maintain over the public Internet because of network delays. Packet loss is another problem. With TCP/IP, lost packets are retransmitted but only after substantial added delay. UDP/IP is an alternative, but UDP does not guarantee packet delivery. So UDP needs a forward error correction scheme.

Pre-standards versions of both systems exist. The TCP/IP approach overcomes delays by a technique called T.30 spoofing. Spoofing slows down protocol interactions with one fax machine to permit the other end to catch up.

I-fax must remain easy to use (no computer required), and provide the deterministic delivery notification typically lacking with Internet e-mail. Eventually, I-fax must become more than just fax over another communication link. Successful I-fax vendors will be those that "integrate" with the e-mail community, and later with multi-media messaging and other emerging Internet technologies.

Migrating Legacy Equipment
There is an immediate market in cutting fax costs by diverting long-distance fax traffic to the Internet or private IP networks. For fax, the legacy telephone network is so ludicrously overpriced that bypass is a no-brainer. Then, as the new I-fax machines are deployed, there will be a huge business opportunity for gateways between legacy fax equipment and the new I-fax machines. There are two basic types of gateways appearing on the market today, store-and-forward and real-time.

Store-And-Forward Gateways: In a store-and-forward gateway, a complete transaction consists of two independent sessions. The originating fax machine sends its document across the PSTN to the gateway as a normal G3 fax session. The image is stored as a TIFF-F file and then transferred as an e-mail attachment though the Internet or other IP network to a remote gateway near the destination. The remote gateway delivers the document to the destination G3 fax machine in a normal G3 fax session over the local phone lines. Store-and-forward typically provides confirmation of delivery via an e-mail or fax return receipt from the destination gateway, although not in real time.

Real-Time Gateways: To work with traditional fax machines, which have page-by-page confirmation, real-time gateways have to meet T.30 time-out constraints. This is easy on the PSTN, but harder on the Internet. There are two strategies for accomplishing this. If the Internet connection is really fast (<1 second delay), you just "demod-remod." Demod-remod means you demodulate the modem tones of the originating fax machine, convert them back into the digital bits, ship these bits over the Internet, remodulate them at a second gateway, and send them over the local voice telephone network at the destination end. Demod-remod works well on high-bandwidth private IP backbones, but not so well on the public Internet.

When the IP network’s delay is greater than 1 second but less than 5 seconds or so, it is possible to use T.30 spoofing as described above. Proprietary T.30 spoofing products are available today, but it will likely be well into 1999 before T.38 is complete and interoperability testing begins.

Within the next year or so, expect universal I-fax gateways to emerge. These gateways will adapt to varying Internet performance. When delays are low, they’ll use demod-remod. As delays increase, they’ll introduce T.30 spoofing. And, if all else fails, they’ll fall back to store-and-forward operation.

THE "FAX" OF THE FUTURE
In the long term, T.37 and T.38 will prove to be transitory specifications. They will provide a decade of interesting and profitable growth for I-fax gateway vendors, but they will eventually be replaced by a richer set of protocols supporting new quality and usability features.

An early indication of what is possible can be seen in the Internet Printing Protocol (IPP), which is currently in development within the IETF’s printer working group. At a sufficient level of abstraction (such as that maintained in this article), the inner-workings of I-fax and IPP are the same. Both involve sending images to remote image-rendering devices and getting back confirmations. Of course, fax is simple, using telephone numbers as addresses. Printers have device drivers — typically a separate one for each destination printer. So for fax to take advantage of IPP, all of the details would have to be hidden in the simpler fax interface.

Indeed, the crux of the problem is not the lower level protocols, but the user interface. Telephones and fax machines use numeric addresses, so a 10-digit keypad suffices. It’s simple — just dial! Internet addresses are alphanumeric — not so simple! I-fax machines can use one-key speed dialing for Internet addresses, but setting up speed dial entries is tedious. Full keyboards are too complex. Look for Palm Pilot-like interfaces and other experiments.

Meanwhile, expect to see telephone numbers assigned for some Internet addresses. The TIPHON project at the European Telecommunications Standards Institute (ETSI) has already asked the ITU to assign a separate telephone "country code" for use with Internet devices.

Eventually, a combination of speech recognition and natural language understanding will make possible a really simple fax user interface. But that’s a different topic (and an article in itself!). For now, expect at least a decade of interesting evolution in the fax market.

Brough Turner is senior vice president of technology at Natural MicroSystems, a leading provider of hardware and software technologies for developers of high-value telecommunications solutions. For more information, call Natural MicroSystems at 508-620-9300 or visit the company's Web site at www.nmss.com.