Fax over IP is alive and kicking

Is fax still relevant in today's e-mail dominated world? G B Kumar, General Manager, Internet Solutions Group and Business Development, Intel thinks fax over IP may be the next big booster for this dying technology

Is fax still relevant in today's e-mail dominated world? Scan through an IP telephony analyst report from the likes of Probe Research or Frost and Sullivan and you'll see that IP fax has tremendous upside and presents a formidable business opportunity. It's not quite there yet, as real time fax today is not part and parcel of a typical Internet telephony service providers’ solution.

Actually, a small percentage of today's IP telephony calls drop off because they connect non-voice tones which the gateways don't understand (meaning a customer is either trying to fax or use a modem). So is fax alive and kicking, or is it at death's door? And what does all this mean for IP telephony?

To answer the first question, let's consider the perspective from a US office. I remember just a few years ago sending and receiving several faxes a day. Today, I still send and receive faxes but now it's only a couple of times a week. Why even that many faxes, you may ask? Because I still send faxes to submit my forms for speaking engagements at trade-shows. Companies like TMC, which sponsors the Internet Telephony Conference and Expo, want my signature to show I'm committed to that event. I also sometimes fax documents to other company offices with my notes scribbled on them, since it's often faster than editing a document online. It's also more convenient to edit a hard copy when travelling. Plus, I frequently receive faxes from my company's offices and customers in the Pacific Rim.

Though e-mail is my primary means of communication, the fact that I'm still using fax drives a point: faxing is still alive and kicking and is important for the point of view of a full IP telephony solution. In this context, we're talking about real-time fax service that fulfills users' expectations of walking up to a fax machine, sending a fax, and having a piece of paper simultaneously come out of a receiving machine at the destination phone number. The bottom line—next-generation service providers and IP centric enterprise communication systems need to ensure that this type of fax capability is part of their infrastructure solution.

Standards
One way of going about implementing fax over IP is by using standards formalized by the ITU (International Telecommunications Union). The ITU has developed two different standards addressing fax over IP. These standards were incorporated in commercial systems in the second half of 1999. The first standard, T.37, is used mainly for store-and-forward faxing. It merely defines the format in which fax is to be delivered as an e-mail attachment. The second standard, T.38, defines the protocol for real-time delivery of fax over IP (FoIP).

With T.37, the fax is sent over IP as e-mail attachment and delivered to the destination over the public switched telephony network (PSTN) by the gateway closest to the destination. While T.37 allows for cost savings through toll arbitrage, from a user's perspective it is a store and forward model and is not real time. You can't immediately receive confirmation that the fax was successfully delivered. Although a sender might claim a fax "has been sent," the T.37 model doesn't let you automatically assume a confirmation of delivery has been sent. If an en route e-mail server happened to be down (perhaps due to some worm or virus attack), the confirmation wont go through.

To satisfy the need for real-time FoIP, it's possible to use the G.711 coder to transmit the fax. The problem here is very high bandwidth usage. Unlike voice transmission where the penalty is just human discomfort; jitter, packet loss or latency can all cause a fax machine to immediately terminate a call. The world is still learning traffic engineering for multimedia transport over IP. In some locations, the bandwidth is extremely limited making network problems virtually impossible to avoid.

Any real-time protocol over IP that bridges the call path must meet requirements of the T.30 protocol (the protocol for standard fax calls over PSTN). The T.30 protocol has very stringent requirements. For example, signals are exchanged every 75ms. To solve this problem, ITU-T Study Group 8 developed the T.38 protocol. Transporting a fax using T.38 takes only a half-duplex channel and 14.4 kbps plus packet overhead. Transporting a fax using the G.711 channel takes a full duplex 64 kbps plus packet overhead.

T.38 is modeled as a smart T.30 interpreter. It executes extensive training, signaling, and data exchange with T.30 to determine the line quality on a PSTN network. This is meaningless with a packet network, since the IP packets can take any available route. Gateways at each end execute full T.30 for communication with fax machines. However, all the data is not transferred over IP. While the fax machine sends the entire CNG/CED-type tones for signaling, the gateways using T.38 only exchange octets that indicate whether they've succeeded or failed at detecting tones.

Essentially, the two gateways only exchange simple results such as confirming success or failure, and only transmit the pages intended for delivery.

The future
The focus is slowly shifting to developing V.34 extensions to protocols in order to allow 33.6 kbps fax transmission over IP. More research is being done in alternative call control models such as SIP and H.248, as well as on standardizing techniques for switching from voice to fax and vice versa for both H.323 and H.248.

So is fax still alive and kicking? Yes. And it will continue to be important for IP telephony. As standards evolve the picture will become clearer.

G B Kumar is General Manager, Internet Solutions Group and Business Development, Intel. The views express in this column are the author's own.