Before the Dawn of IP Telephony - Part 24Develop a broadband VoIP gateway (June ~ October 2001)
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
Due to the drastic drop in prices of broadband services, we started the development of a broadband VoIP gateway. It was a development effort with different requirements from the corporate gateway.
Photo 1 : VA14, first broadband VoIP gateway
The development of "VA14," was the first VoIP gateway for broadband network providers (photo 1). There were two issues: Price and delivery. We were expected to achieve short delivery at a low price.
When thinking in terms of cost reduction, use of LSI had the potential for drastically reducing the price. It was also a tradition at OKI to pursue what is ideal in terms of technology and the development of a special LSI was an option. However, we were still unclear on the market scale at that time and there was no guarantee that sales would cover the development cost.
Furthermore, we would never make it in time for commercialization in 2001 if an LSI was developed at the normal speed. We needed a new approach to accelerate development.
Keeping down material cost with common parts
Material cost drop as more and more are used. If the cost cannot be reduced, the quantity cannot be increased and thus there exists a negative cycle. Price drop, however, when the quantity starts to increase triggered by some event. When the price drop, the quantity further increases, creating a positive cycle. It is the paradox of "what comes first: the chicken or the egg?" The most important factor in this product development was to create a "trigger" for this positive cycle. But what?
We found the answer in common parts. There was not enough time for creating a special LSI and it was difficult to have a grasp on an unclear demand. So, we decided to thoroughly use common parts that can be found in the market in huge quantities even if slightly redundant in terms of features. These, for example, are parts used in cellular telephones.
Although this measure did not lead to the ultimate device cost cutback since it steered away from the ideal design of developing a special LSI, cost would certainly decrease. And I gave this certainty priority.
There is the viewpoint that use of special parts will lead to an ideal and most efficient design. However, I chose to use materials that were out in the market in large quantities rather than designing ideal, special parts since I felt it would be a quicker route when creating something so unclear in the period of uncertainty. When a product is released, opening a new market, there will be competition of developing special parts for the product and results in cutback of material cost. In other words, the possibility of creating in large quantities increases when decrease in material cost can be expected. Then, material cost decreases even further and enters a positive cycle. In such way, I first decided to use common parts and hoped for a positive cycle.
Use of materials that can be found in large quantities also works to an advantage when procuring materials. Since there is high possibility that materials can be found even when there is an increase in production volume, production lead time can be shortened. In addition, there is also the advantage of stable reliability since materials already with results in the market are used and quality of hardware can be maintained to a certain level.
Problem is conversion of hardware to software
Although everything seemed to work at our advantage, there is a major problem in the development of products using such common materials. It is the relocation of function assignment. When structuring a product using common materials, the function assignment will not always be ideal. For this reason, the pattern of function assignment structured by used materials must be relocated. For example, there may be the need to use software processing for something that was processed using hardware. This means using software to realize what was being achieved with hardware codec in terms of voice quality.
Normally, hardware engineers and software engineers have different roles. In order to create what was being achieved with hardware using software requires filling of a huge gap both technically and consciously. Creating something that was achieved by hardware up to then with software is more difficult than one might assume.
Key to success is to create all elemental technologies yourself
Despite the difficulty of turning hardware into software, we were able to accomplish the task. We believe there are two reasons for the success.
The first is that we had created all VoIP elemental technologies ourselves. Hence, we were fully aware of the systems and functions that were to be realized. In other words, there was no black box. For the required functions, we used technologies capable of creating the solution using hardware or software, enabling a smooth conversion of functions, using software instead of hardware. From signal process of VoIP to system development, we believed that our honest efforts had paid off.
The second is that the development organization of the project consisted of both hardware and software engineers. It was at the time when the concept of technical integration was being adopted within the company. As a result, we were able to flexibly assign roles for the project without being particular about the role assignment of the organization.
The development of "VA14" was completed in October 2001. It was an incredibly fast development of three months since launch.
... To be continued
