The "synchronous digital hierarchy" (SDH) and the "asynchronous transfer mode" (ATM) are fundamental to the emerging broadband ISDN. Among other attributes, SDH, a version of which is called Sonet in the United States, for the first time internationally standardizes bit rates and transmission framing structures, simplifying networking between countries. The lowest bit rates (155.52 Mbit/s for SDH and 51.84 Mbit/s for Sonet) are adequate for all broadband services, including high-speed data and computer connections, video, and high-resolution graphics. Since "broadband" brings with it new services requiring new bit rates, and because these services may have widely differing traffic characteristics as well as switching requirements, ATM was devised and standardized to accommodate all these different needs simultaneously. ATM can multiplex voice, data, and video traffic, both packet-switched and circuit-switched. Furthermore, ATM switches can as readily switch any of these services as any other. This is unlike today's switches, which are optimized for particular bit rates and services.
NTT is strongly committed to both SDH and ATM. Because SDH and ATM are new international standards for telecommunications, products developed for domestic use have immediate worldwide marketability, although systems used in North American, Europe, and Japan are not identical. Japan and North American currently use the same hierarchy up through 6.312 Mbit/s, but diverge above that. Europe differs from Japan and North America at all levels above 64 kbit/s. Different transmission standards in the three regions necessitate different products for each region. Companies such as NEC and Fujitsu already have a strong presence in transmission systems in the United States, and the JTEC panel found other companies such as Oki, Hitachi, and Toshiba to be strong contenders for future consideration in this area.
At the Fujitsu plant in Yamanashi, the panel saw the components for Fujitsu's commercial 2.5 Gbit/s SDH long-haul trunk transmission system. These consisted of eight gallium arsenide integrated circuits (GaAs ICs) for all multiplexing, transmitting, and receiving functions, plus the distributed-feedback (DFB) laser transmitter, the avalanche-photodetector (APD) receiver (-32 dBm sensitivity, gain <10), and an EDFA pumped at 1480 nm. the system operates to 200 km. Fujitsu's Yamanashi plant is currently developing the next-generation 10 Gbit/s system.
At Fujitsu Labs in Atsugi, the panel saw a multi-quantum-well (MQW) DFB laser with modulator that is intended for 1.55 microns, 10 Gbit/s upgrades in 1.3 Ám transmission systems that use previously installed conventional single-mode fiber. Chirp from a directly modulated laser at 1.55 microns, where fiber dispersion is approximately 17 ps/nm/km, would induce severe penalties and reduce span lengths, canceling the advantage of lower attenuation at 1.55 microns. Solutions include combining a narrow-linewidth laser with an external modulator (plus isolator to prevent reflections from reentering the DFB), or integrating the modulator with the laser. Fujitsu integrates a strained-layer MQW-active-region DFB laser (10 mA room-temperature threshold) with an MQW absorption modulator having a 3 dB bandwidth of 12 GHz. This combination introduces only a 3 dB loss relative to the 8 dB expected using an external modulator. The lab has performed 70 km transmission experiments at 10 mW and 10 Gbit/s and found negligible dispersion penalties. At higher power levels (narrower linewidths), distances up to 120 km are expected.
Oki currently manufactures 156 Mbit/s and 2.5 Gbit/s SDH systems, is developing 10 Gbit/s systems, and is in various stages of R&D on 40 Gbit/s and 160 Gbit/s systems. For their 10 Gbit/s product, Oki researchers are designing around 0.3 microns GaAs MESFET (metal-semiconductor field-effect transistor) technology for the highest-speed chips and 0.5 microns MESFETs for lower-speed sections (622 Mbit/s). The transmitter options being considered, again, are a narrow-linewidth MQW laser combined with either an external LiNbO 3 Mach-Zehnder modulator, or with an integrated electroabsorption modulator. Oki favors the latter option and has it under development to be ready in two years. The receiver is based on a pin photodiode plus GaAs IC preamp, combined with an EDFA as an optical preamp. This raises the sensitivity from -15 dBm to -33 dBm at a bit-error rate or ratio (BER) of 10-11. Oki is currently building a 10-Gbit/s system prototype for NTT, and along with Hitachi, Fujitsu, and NEC, is producing EDFAs for NTT for 10 Gbit/s systems.
The JTEC panel's hosts at Hitachi described their approach to transmission beyond 10 Gbit/s, which they believe will suffice until the year 2000. Beyond 2000, they expect to use four wavelengths, each modulated at 10 Gbit/s, to reach 40 Gbit/s, and, beyond the year 2005, to use eight wavelengths, each modulated at 20 Gbit/s, to reach 160 Gbit/s.
Hitachi, Fujitsu, and NEC currently have a strong presence in ATM switches, and undoubtedly Oki, Toshiba, and others are potential suppliers. Oki and Fujitsu are both carrying out research on optical ATM switches, and in a short movie, Toshiba showed panelists a 64 x 64 Batcher-banyan-like ATM switch operating at 155 Mbit/s and a 10 Gbit/s long-haul transmission system it has developed.
In the area of laser development, optical amplifiers, and associated electronic integrated circuits for 10 Gbit/s (and higher) backbone networks (because of NTT's emphasis on this area), Japanese manufacturers are ahead of the United States. In ATM switches, an area of intense development by small as well as large companies in the United States for local-area networking as well as telephone switching, Japan and the United States are approximately even as far as JTEC panelists could tell.