The Internet is an indispensable part of our lives. The amount of communication through the Internet is increasing at an annual rate of 40 %. Can it continue to increase forever? Fig. 1 shows the future estimation of network communication amount and power consumption of routers. The black and red dots are based on government survey data, and the actual lines are extrapolated from the observed annual increase rates. What can be seen from the figure is that the total router power consumption increases as the communication volume increases, and it took up about 1 % of the total electric power supply in 2006. This means that to enable 100-times increase of communication load in the future, the power consumption of the router needs to become 100 % of the total electric power supply. The router, of course, will become more highly efficient in the future, but nevertheless, the present technology will face limit.

On the other hand, looking at the transition of network related technology as the cause of increase in communication volume, one can see that the use of network has moved from the telephone (voice) to data communication, and is presently shifting toward video contents (Fig. 2). It is thought that super high-definition imaging technology such as super hi-vision^[1] will be widely used in the future. Then, the concept of "remote co-existence" through the network will become possible, and most of the activities that were done by meeting in person will be able to be done at a distance. For example, 24 hour sharing of space information through a screen which occupies the entire wall of a living room will make "distant living together" situation become technologically feasible.

We need to create a network technology that can handle volumes larger by several digits with which we can easily exchange high definition videos, without increasing energy consumption. Looking at Fig. 2, one can see that the granularity per user (unit of data volume processed by a user through the network) is changing from digital packet to optical fiber line (or optical fiber path). Keeping this in mind, we are focusing on optical path network or an optical circuit switching system which switches optical fibers/paths. As can be seen in Fig. 3, compared to existing packet switching, if only considering node processing, optical path network which switches optical fibers/paths with optical switches can save energy by 3, 4 digits.

There is no other low-power technology that is as simple and that has large potential as this. However, optical path network is not a packet switching network as the existing one but a circuit switching one. Fig. 4 shows the difference between data transfer between computers, and transfer of a live image of a person. Simply said, packet switching is suitable for the former, whereas circuit switching is suitable for the latter. As the network handles both data and videos, the packet switching and the circuit switching are complementary. Therefore, we think that it is appropriate to do research and development to establish an optical path network parallel to the existing packet switching networks. Fig. 5 shows the total view of the framework concept and the enabling technologies.

The key enabling technologies for optical path network such as optical switch, high-speed interface, optical amplifier and dispersion-managed transmission line have already been put to practical use. However, to use optical path network "anytime, anywhere, freely, by anyone", a massive deployment of fiber will be necessary, and the existing hardware technology is almost totally useless in scale, volume, cost and size. A new technology is needed which will enable further integration, mass production and cost reduction in fundamental devices. Moreover, even if the hardware technology matures, there is a need to develop various kinds of software depending on how the network is going to be used.

It is vital to look over from applications to fundamental devices and discuss/envision the best new network architecture, and to set the goal of individual enabling technology. In this context, especially related to optical technology are optical path interface technology such as network interface card (NIC) which physically connects optical path and applications, optical path management and control technology such as middleware which autonomously optimizes the optical path settings upon users’ requests, optical path switching technology such as optical switch which physically switches over optical paths, and optical path conditioning technology such as an autonomously controlled tunable dispersion compensation device which enables optical signals to be transmitted correctly when optical path switches. AIST, while aiming to become a "vertically integrated center for optical path network technologies" in collaboration with external companies and research institutes, will continue to do research and development of these enabling technologies^[2].