Before the Dawn of IP Telephony - Part 4The difficulties of convincing (latter half of 1996)
These contents translated a serialization article carried by ITPro IP telephony ONLINE published by Nikkei Business Publications, Inc. Jump to the original (Japanese).
Shinji Usuba
General Manager
eSound Venture Unit
Oki Electric Industry Co., Ltd
In April 1996, VOICEHUB that converts voice to IP packets was displayed as reference exhibition at Communication Tokyo for the first time. The success led to the use of VOICEHUB in a system by a certain integrator. Despite the exciting news, there was only seven months to shipment. And we also had to convince the integrator the device really works.
Photo 1 : OKI Annex 6 located in Shibaura, Tokyo
Although the use of VOICEHUB was decided, we heard rumors that the integrator was extremely skeptical about converting voice to IP packets. And they had every right to be concerned. The device was based on an entirely new concept at the time. We would cause inconvenience to customers if we could not establish sufficient quality that could endure practical application. In order to send the device out into the world, we made frequent visits to the integrator. The integrator members also visited OKI on a number of occasions (photo 1).
From the very beginning, the idea of voice packets was not embraced by the traditional telecom industry. But the integrator was very patient and did not reject the idea without giving us the benefit of the doubt to answer their questions.
Isn't there start-up distortion? How is the latency? Are echoes noticeable? Are there voice intermittence? Their questions aimed directly at the weakness of voice packets. Although it took some convincing, such questions were very useful for brushing-up VOICEHUB from the prototype level.
Such exchange went on for about three months to bridge the gap between both parties. Since there was no way of showing the performance history of stable operation, all we could do was to explain the mechanism behind the prototype. Although the members in charge of sales were persistent, the integrator was equally patient with us.
The last remaining issue was "latency caused by jitter." A jitter refers to the variation in time each packet takes to reach the other party. When sorting the traffic of data passing over an IP network, jitter occurs when there is wait time. Voice intermittence occurs when this wait time becomes too long.
IP networks at the time had major jitters, mostly due to performance of devices not being good as the present. Although it was the inevitable fate of best-effort communication where speed cannot be guaranteed, we somehow had to overcome this problem.
When using a large receive buffer in VOICEHUB to absorb jittering, latency occurs as the time it takes for the voice to reach the other party becomes longer. This also inhibits natural communication.
On the other hand, smaller receive buffer means less latency, but voice intermittence from jittering would occur. This cannot be tolerated for practical use. Solving this conflicting issue was the key to convince the integrator.
Time passed relentlessly. And we were gradually nearing the time limit for completing the design of the mass production model.
The crucial point
At the beginning of October 1996, five members at OKI - who were desperate to release the product into the market - and the key person on the integrator side to decide whether VOICEHUB was of any use met at the integrator's office in Tokyo. This was the final day when it was to be decided whether or not VOICEHUB would be used.
I was confident that the VOICEHUB can maintain voice quality. If we could just convince the integrator on the details, we would surely get the go sign. If the answer is no, then we had to come up with an alternate plan.
Is voice interrupted when exceeding the buffer size? What would happen if there's no jittering latency? What would happen if the jittering latency is, for example, 700msec? What happens when there is packet loss? Questions flew one after another. And I drew diagrams to answer each one.
And then, the face of the key person on the integrator side turned to a smile. At that moment I knew that the problem had been solved. We managed to overcome the brick wall. It was a sign of relief. We finally convinced the integrator the system would prevent unnecessary latency and voice interruption.
The integrator had thrown his doubts at us because he was serious about using VOICEHUB. Although the exchange was difficult, it boosted our confidence in releasing the product into the market.
And the exchange helped us to create the VoIP gateway market and commercialization of high-quality VoIP gateway products not dependent on network environments. We were extremely grateful.
At the beginning of October 1996, the device specifications were officially decided. Functions, specifications and price were determined. At this point, I thought that all we needed to do was to head towards completion of the product that would meet the quality, cost and delivery conditions. Of course, things are never that easy.
... To be continued
