ECOC: Pick of the Papers

AMSTERDAM -- ECOC 2001 -- The current slump in the tech industry hasn't dealt a killer blow to innovation -- it's just made it harder to find.

Plenty of exciting developments are coming out of research labs around the world, according to the postdeadline papers presented at the technical conference linked to the ECOC Exhibition.

The postdeadline papers represent the latest and most significant developments in optical technology. Lots of researchers submit papers on their most recent achievements, and only a small proportion of them end up being accepted by panels of recognized experts.

What follows is Light Reading's take on the papers that generated the most excitement. For those wanting to dig deeper, the research paper reference numbers [in brackets] are provided.

Hero Experiments: Faster, Wider, Longer

No fewer than a dozen postdeadline papers detailed ways of improving capacity and reach in the optical network.

While the 10-Tbit/s barrier was broken earlier this year by two companies -- Alcatel SA (NYSE: ALA; Paris: CGEP:PA), and NEC Corp. (Nasdaq: NIPNY) -- the transmission distance involved was very short, less than 120 km (see Alcatel Holds World Record for a Day). Results presented at ECOC focused on increasing reach, but at the expense of some capacity.

  • This included the same team from Alcatel, which managed to send 5.12 Tbit/s over 1,200 km of Teralight fiber without electrical regeneration [PD.M.1.1].
  • Also, TyCom Ltd. (NYSE: TCM; BSX: TCM) sent 1.28 Tbit/s over 4,500 km of fiber [PD.M.1.2]. Both experiments used 40-Gbit/s channels but differ in numerous other ways. One way to compare them is to consider the capacity/distance product, which in both cases is around 6,000 Tbit/s per kilometer.

  • Another way of measuring capacity advances is to quote spectral efficiency, which characterizes how many high-capacity channels can be squeezed into a particular wavelength band. Japan's government laboratory CRL and Osaka University claimed a record efficiency of 1.6 bit/s/Hz of spectrum in their experiment, which was unusual in that it exploited optical code division multiplexing (OCDM) [PD.M.1.3]. (The previous best was held by Alcatel in the 10.2-Tbit/s experiment mentioned above.) The Japanese team achieved a total capacity of 6.4 Tbit/s using just 40 wavelengths, but the transmission distance was only 80 km.

  • Other notable results include Alcatel's use of 25 GHz channel spacings in a 380 km unrepeatered link [PD.M.1.5]. In addition Agere Systems (NYSE: AGR) reported single-channel 160-Gbit/s transmission over 200 km of non-zero-dispersion-shifted fiber [PD.M.1.10].

    Optical Packet Switching

    The general consensus is that packet switches will need to be all-optical to achieve terabit-per-second throughputs, but significant hurdles remain in the way, not least of which is the fact that so far there is no method for optically storing packets (see Storing Light). But the lack of an optical equivalent to electronic memory is not stopping researchers from building prototype optical packet switches in the lab.

    The usual suspects are behind the latest breakthroughs in optical packet switching: Alcatel, Bell Labs, and Japan's CRL/Osaka University.

  • The Japanese team deserves first mention, as it has developed a system that truly has no electronic data processing [PD.A.1.9]. Instead, the packet switch is based around an all-optical header lookup, which decides whether or not a packet can pass at a node. The switch actually achieves terabit capacity by carrying 160 Gbit/s on each of 8 channels. But it's going to be some time before the role of 160 Gbit/s in real-world networks becomes clear.

  • Other prototypes from Alcatel and Bell Labs are intermediate between electrical and optical. Packet headers are still read electronically, and contention issues resolved electronically. But this all happens on the edge of the equipment.

  • Bell Labs' scheme deploys tunable lasers and an NxN arrayed waveguide grating (AWG) [PD.A.1.7].
  • Alcatel's innovation uses a broadcast-and-select scheme, which avoids the need for tunable lasers [PD.A.1.8]. In addition, Alcatel demonstrates an asynchronous optical receiver -- "something that is often overlooked in previous works, although it is essential to such a system," according to the researchers.

    All-Optical Signal Processing The basic idea behind all-optical signal processing is to replace electronic gear with protocol and bit-rate transparent optical equipment. All-optical 3R regeneration (so-called because it reamplifies, retimes, and reshapes the signal), optical demultiplexing, wavelength conversion, and signal processing tasks such as optical reading of packet headers all fall in this camp. It's billed as increasingly important, if not essential, as networks reach speeds of 40 Gbit/s and beyond.

  • Siemens AG (NYSE: SI; Frankfurt: SIE) reported virtually error-free demultiplexing from 160 Gbit/s to 10 Gbit/s using a Mach-Zehnder interferometer incorporating semiconductor optical amplifiers (SOAs) [PD.B.1.8]. This differed from earlier results in that the device was monolithically integrated.

  • A key part of the optical 3R regeneration process is to implement bit-rate flexible clock recovery. Researchers from the Heinrich Hertz Institute reported a method of extracting 25 to 82 GHz clock signals, which is fast enough to retime a 160 GHz NRZ (non-return to zero) signal [Th.F.1.2].

  • The University of Paderborn, Germany, reported efficient single and multichannel wavelength conversion using difference frequency generation in periodically-poled lithium niobate [PD.F.1.10]. The wavelength converted signal was then transmitted over 500 km of fiber.

    Planar Integrated Circuits The drive to miniaturize optical subsystems using waveguides continues. Here are some highlights:

  • An array of MEMS (micro-electro-mechanical system) mirrors is tiny, yet Lucent's Lambdarouter occupies many equipment racks. Why? All the additional space is occupied by the components surrounding the switch matrix, such as multiplexers, splitters, amplifiers and so on. Now, Agere and Bell Labs report a compact 72x72-port crossconnect that incorporates a stack of AWGs that interface directly to a MEMS switch fabric [PD.B.1.5].

  • Telephotonics Inc. reported a 16x16 switch matrix based on a polymer chip measuring 4x10.4 cm. The power consumption was just 6.4W, and the optical losses totalled 6 dB [PD.B.1.6].

  • Serious progress was made in waveguides based on photonic crystals, which are defined by a repeating pattern of deep-etched holes. Scientists at the Royal Institute of Technology, in Kista, Sweden, reported indium-phosphide-based photonic crystal waveguides with low losses: The best case was as low as 1 dB/100 microns [PD.F.1.5].

    — Pauline Rigby, Senior Editor, Light Reading
  • Be the first to post a comment regarding this story.
    Sign In