Before the Dawn of IP Telephony - Part 1Uphill battle for the development of the first IP telephony (1995 ~ 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
Handling voice over an IP network-a concept that seems somewhat of common sense at the present with the widespread of IP Telephony was a challenge that was simply unheard of 10 years ago. It was about then when I first started the development of IP telephony related products. Here's a review of the efforts at the time.
Photo 1 : OKI's booth at NET&COM 2004
Half of the display booth was dominated by eSound IP phones, realizing voice quality that surpasses conventional telephones.
At "NET&COM 2004," an exhibition held in February 2004 in Makuhari, Chiba, OKI displayed the eSound IP phones, an IP telephony system for offices (photo 1).
eSound IP phone is a strategic product aiming at voice quality that exceeds that of conventional subscriber telephones. The objective was not only to sweep away the preconception of "IP phones have bad sound quality," but to go further beyond. Fortunately, many visitors tested and praised the superior voice quality of the eSound IP phone. Watching the visitors, I was hit with a sense of nostalgia as I have been working on IP phones literally since incunabulum.
The road leading to the present was never easy. Standing at the booth, I had recalled memories of "VOICEHUB," the first IP telephony related product that I developed 10 years ago.
"Create a device that handles voice over LAN"
It was a day in August 1995 when I was standing blankly in front of my supervisor. He wanted me to examine whether we could create a system that handles voice over LAN.
Handle voice over LAN? The concept may seem nothing out of the ordinary at the present with the widespread of personal IP telephony services; however, the request exceeded the boundaries of common sense for me at the time. No, it was a request that would have exceeded the boundaries of common sense for everyone-inside the company and out-not just me.
In order to handle voice over LAN, it had to be converted to packets in the same manner as computer communication. There have been researches on voice packet technology, and it was known that with IP communication, packets at times do not reach the other party. This led to voice intermittence. Unlike data, voice simply cannot be resent. Furthermore, voice packets being researched at the time were based on the concept of sending only portions present with sound during the actual communication, in order to save used bandwidth. Detecting this presence of sound is difficult, leading to a phenomenon called "start-up distortion" where the beginning of a conversation is cut-off and not played back properly. Practical application of voice packets was considered virtually impossible.
For this reason, another technology referred to as ATM (Asynchronous Transfer Mode), built with a system that maintains quality of voice communication, was assumed to become the future trend. ATM is a method that sends data, such as voice, by converting them into packets of fixed length. The technology enables communication to be processed using hardware to minimize packet loss and latency. Many thought that voice and data can be integrated using ATM.
As I stood blankly, my supervisor continued, "We are entering the PC era. The LAN device market will also expand. In the not-too-distant future, prices will drop and what was available only as commercial use will be opened to the general public." His tone became firm, "This is where 'voice' comes in. Voice traffic is overwhelmingly smaller than data traffic. I assure you: there will come a day when voice is integrated with data network including the Internet."
Hearing the words of my supervisor, so full of confidence, I felt a sense of motivation. If we could integrate voice and data network, we may be able to open a new world of communication.
At the time, I was in charge of PBX (Private Branch eXchange) development. It was just when I was feeling stagnated since development of new functions had taken on a loop and I couldn't see what was coming next. I was also stimulated by the fact that supervisors at the time were promoting exchange among telephone system engineers and computer engineers.
"Let me do it," the words spilled out without a moment of hesitation. The words reflected my true and honest opinion.
This marked the beginning of my involvement in the development of VOICEHUB, the pioneer of IP telephony related products at OKI. The goal was to display a working device at "Communication Tokyo," a general exhibition of communication devices to be held in April 1996.
Frequent visits to Takasaki to learn computers
The project was launched immediately and members of the department were gathered. There were 8 members on the team, all young that have joined OKI within the last 6 years. Everyone was extremely aware that they were pioneering into a new territory.
Photo 2 : "Micro Card 450-10," a card-type PC developed in August 1995 by OKI
Credit-card sized board built with PC functions equivalent to an IBM compatible machine. The card can be used as parts of other products.
It wasn't too soon that we had hit a brick wall. Computer knowledge was indispensable in the development of IP telephony products. But no member had knowledge about computers. Consistently working in the department developing communication devices, I had confidence in telephone and voice technologies. But no knowledge concerning computers such as IP networks and LAN. Even if I managed to create voice packets, I did not posses the skill to effectively apply Internet protocols necessary for handling voice on an IP network.
But there was no time for newly recruiting members from the Computer Department. We were devastated. What saved us was the in-house exhibition. On display was a card-type PC (photo 2). Then it hit me. We should be able to create the ideal device if we can fully utilize this card-type PC.
The card-type PC was created by the Design Department in charge of information-type products. At the time, the department was based in Takasaki of Gunma Prefecture. The department was remote in terms of both fields of specialty and geographical location from our Communications Department located within Tokyo.
Photo 3 : OKI's Takasaki Office, where Usuba frequently visited for product development
But there was there was no time to complain. Despite my impudence, I frequently visited a section manager at Takasaki (photo 3), who was in charge of the card-type PC design. He filled me with passion and even offered the development environment of information-type products. As a result, the image of VOICEHUB had taken shape quickly.
Fatal error discovered 1 month prior to exhibition
The actual device was developed from the ground up. Combination of hardware, software and LSI (Large Scale Integration) circuits were also new. The development seemed to seesaw, sometimes advancing and sometimes retreating.
The exhibition, which we had considered to be the goal, was only a month away by the time all modules according to the first design were ready. When the prototype was finally assembled, I felt a sense of relief. Immediately, we gave it a try. No sound.
As a result of verification, we found that voice packets were sent successfully. But we discovered that there was an error in the process where the device on the receiving-side restored the packets into the original voice. At first, we thought it was a configuration error. But there were no obvious errors. Why is there no sound?
What we soon discovered was a fatal error. "It's an LSI bug."
My jaw literally dropped when I heard the words. LSI bugs are commonly known to be fatal. We called the LSI developer and examined possible measures, but there seemed to be no easy solution. And there certainly was no time to develop the LSI again from scratch.
At this stage, where in the booth of the exhibition VOICEHUB would be placed was already determined. Leaflets were ready. I started to lose my confidence. Maybe we should give up the idea of displaying a working device?
It was Mr. Kuboki, one of my subordinates, who spoke up while my head was buried in my hands. "We may still have a chance..."
... To be continued
About the author
Shinji Usuba involves himself in the development of IP telephony related products from 1995. More than 10 models have been released to date. His years in development work come to an end when he decides to take on new challenges after participating in an in-house entrepreneur training program. He is now the general manager of "eSound Venture Unit" and responsible for product planning. He is actively pursuing improving IP products with superior sound.
His perfect happiness is to relax in a hot spring that is not too hot. This is also where many of his ideas come from. He has been in a good mood lately as there have been many natural hot springs opening within the city. His favorite is Jindaiji hot spring in Chofu. His family is starting to become tired as he spends much time taking them to the hot springs as an excuse to be with the family.
