How Version 3 of T.38 Increased Internet Fax Speed
Economic conditions being what they are, the Fax over IP (FoIP) user community is looking for an edge. V.34 ‘Super G3′ fax, covered by version 3 of T.38, makes Internet fax transmissions really move, saving money and bandwidth. The original version 0 of T.38 was approved over 15 years ago and, with updates that have gone into version 2, became a mainstay of FoIP-capable gateway designs. Version 3 brought compatibility with V.34 capable fax machines that are now both commonly available and reliable.
Wait…You Can’t Plug Fax into VoIP Gateways?
Much of T.38′s prominence has come about because Voice over IP (VoIP) revolutionized voice communications and became a must-have aspect of corporate telecom installations. It was purveyed as the 21st Century replacement for the old analog PSTN switched circuit telephone system that had been around for close to a century. However, the rapidly expanding VoIP user community took the concept of ‘PSTN replacement’ literally and wanted to plug their fax machines into their VoIP gateways. After all, voice is sound and fax transmissions are sound. Shouldn’t fax work just as well as voice over these wonderful new gateways? Reactions varied from dismay to outrage when the answer came back as ‘no, not really.’
The problem was that, while VoIP could maintain a very respectable quality of service even with dropped packets right and left, packet loss and delay would really clobber fax transmissions over the same system. A wide array of timing considerations and signal compensations had been built into the T.30 fax protocol for what now seems an antiquated analog phone system. With the coming of VoIP, they caused genuine grief when fax signals were pushed as simple digitized packets across an IP network.
Using G.711 for Fax over IP
And this was the initial solution for Fax over IP. Digitize the fax modulation symbols, basically audio carrier tones deviated in frequency and amplitude, stuff ‘em into packets and send them on to their destinations. This simplistic digitize, transmit, re-integrate approach to fax was carried by the G.711 transport protocol which had enjoyed considerable success in voice transmissions. Unfortunately, as soon as a packet was lost, delayed or arrived out of order, the fax call was dead on arrival and G.711 offered no protection from this. And, the really bad news was that a dropped call had to be re-dialed.
This meant that, as fax calls were piled into queues for IP transmission, stopping to re-dial those that were dropped caused these queues to back up, delaying this ‘super speedy’ transmission system more and more. The obvious solution was to take the T.30 fax signals and turn them into specific IP signals that were adapted to the switched packet transmission medium, and this is exactly what T.38 does. Each type of T.30 signal has an equivalent T.38 message that can be easily transferred from the emitting gateway (originator terminal connection) to the receiving gateway (answerer terminal connection). The packets are numbered, the gateways are programmed to reassemble them in the right order according to their function and much of the Internet induced fax grief is cured… but not all of it.
Some of the timing considerations still posed challenges for FoIP even with T.38, but the worst of it was that here was this network with awesome data transfer capabilities saddled with plodding slow fax modulation data rates. A fax is not going anywhere faster than the analog fax terminal can pump it into its gateway. Commercial users pay for telecom services based on data rates and bandwidth. More bandwidth costs more, as do longer transmission durations caused by slow data rates. Once again the VoIP users were becoming restless. They could use V.34 over the PSTN but its bandwidth limitations constrained data throughput to something less than the advertised 33,600 bps that V.34 is justifiably famous for. And there was the Internet, more than capable of handling the V.34 data rates but T.38 stopped at V.17 (14,400bps) and V.34 fax calls would be ‘passed through’ the IP gateway links using… you guessed it, G.711.
Adapt T.38 to Handle V.34 Fax
As before, the solution was obvious: adapt T.38 to handle V.34 fax. Thus, version 3 of T.38 was born just as the V.34 fax terminals really got their interoperability act together and took high speed fax on the road. This fortuitous coincidence was unfortunately stymied because there were no V3 T.38 gateways available to move these much higher speed transmissions across the Internet. VoIP gateway manufacturers had struggled mightily to get their gateways up to V2 compatibility with T.38 and they had spent years listening to the fax terminal manufacturers decry the problems with their implementations of the V.34 modulation.
And, for all the fact that a gateway’s output is IP packets and not paper, it would still need an analog V.34 implementation to take advantage of V3 of T.38. They can be forgiven for being reluctant to take the pain and strain of V.34 development on their own backs. Alas, the telecom market is not a very forgiving place. Corporations with now massive investments in VoIP installations wanted the bandwidth and transmission time savings from V.34 fax… and they wanted them now.
The gateway manufactures heard the call and stepped up to the plate. Version 3 of T.38 carrying V.34 fax calls is now available on a wide array of Fax over IP equipment. The stopper has now been yanked out of the genie’s bottle and V3 T.38 is on the check list of every company that is putting VoIP gear on its year-end shopping list.
Do you disagree? Tell us about it.