Developments in Wavelength Division Multiple Access Networking

0163-6804/98/$10.00 © 1998 IEEE avelength-division multiplexing (WDM) as a system concept goes back to the early 1970s. At that time, however, it usually only implied the use of two wavelengths, generally one in each of the 1300 nm and 1550 nm windows to enable full duplex transmission. Furthermore, the development of point-to-point links was the main thrust of WDM research throughout the 1970s and early 1980s. An early experimental broadcast and select network, however, was demonstrated by Toshiba in 1985 [1]. This network used light emitting diodes (LEDs) to distribute fourchannel analog video. A following significant stage in the development of WDM networks involved the LAMBDANET project [2], developed by Bellcore in 1987. This was an 18wavelength system, which used distributed feedback (DFB) lasers to broadcast to all the network nodes via a passive star coupler. Each node had its own unique wavelength to transmit, and an array of receivers which could detect all the wavelengths on the network. In the same year the concept of the multihop network [1] was proposed to overcome the need for tunablity in either or both of the transmitters or receivers. A number of theoretical studies have been carried out which have resulted in a number of variants to that original Shufflenet proposal. The years 1987–1990 saw the development of the passive photonic loop, a double star network which used static wavelength routing to transmit from a central office to subscriber nodes, and also the development of a number of photonic switches, such as the photonic knockout switch, Fox/hypass/bhypass, and the star-track network [2]. Although the latter were not strictly networks, the architectural and protocol design issues were very similar. A significant development then occurred in 1991 with the arrival of commercially available tunable filters. This enabled the use of tunable receivers instead of arrays of fixed receivers. IBM took advantage of this technology in their Rainbow projects [3]. The realization of the erbium doped fiber amplifier (EDFA), developed in the late 1980s and in extensive use since 1990, was a major boost to WDM systems, since it allowed the amplification of all the channels in a fiber simultaneously. This technology enabled WDM networks to be expanded, in terms of both distance between nodes and the number of nodes. The first major experimental investigation of a multihop network, the Teranet, was undertaken by Columbia University in 1991 [2]. This trial demonstrated 1 Gb/s ATM packet switching and 1 Gb/s circuit switching. Another multihop network, the Starnet, was implemented at Stanford University in 1993 [4]. Meanwhile, development of the single-hop system continued with the second IBM Rainbow project (1994) and the use of a broadcast-and-select network in the outer layer of the All Optical Network (AON) project [5]. More recent developments have included interest in the use of subcarrier multiplexing for control purposes, as in the 1995 CORD [3] and 1996 Starnet II [6] projects, and single-hop ring networks. The latter have often been proposed, but only in the last two years or so have they been experimentally implemented [7, 8].