counter for iweb
Silicon Photonics

Published book, click here

« Fibre-to-the-FPGA | Main | Chinese optical component vendors set for change »

Terabit Consortium embraces OFDM

A project to develop optical networks using terabit light paths has been announced by a consortium of Israeli companies and universities. The Tera Santa Consortium will spend 3-5 years developing orthogonal frequency division multiplexing (OFDM)-based terabit optical networking equipment.


“This project is very challenging and very important”

Shai Stein, Tera Santa Consortium





Given the continual growth in IP traffic, higher-speed light paths are going to be needed, says Shai Stein, chairman of the Tera Santa Consortium and ECI Telecom’s CTO: “If 100 Gigabit is starting to be deployed, within five years we’ll start to see links with tenfold that capacity, meaning one Terabit.”

The project is funded by the seven participating firms and the Israeli Government. According to Stern, the Government has invested little in optical projects in recent years. “When we look at the [Israeli] academies and industry capabilities in optical, there is no justification for this,” says Stern. “We went with this initiative in order to get Government funding for something very challenging that will position us in a totally different place worldwide.”


Orthogonal frequency division multiplexing

OFDM differs from traditional dense wavelength division multiplexing (DWDM) technology in how fibre bandwidth is used. Rather than sending all the information on a lightpath within a single 50 or 100GHz channel – dubbed single-carrier transmission – OFDM uses multiple narrow carriers.  “Instead of using the whole bandwidth in one bulk and transmitting the information over it, [with OFDM] you divide the spectrum into pieces and on each you transmit a portion of the data,” says Stein. “Each sub-carrier is very narrow and the summation of all of them is the transmission.”

“Each time there is a new arena in telecom we find that there is a battle between single carrier modulation and OFDM; VDSL began as single carrier and later moved to OFDM,” says Amitai Melamed, involved in the project and a member of ECI’s CTO office. “In the optical domain, before running to [use] single-carrier modulation as is currently done at 100 Gigabit, it is better to look at the OFDM domain in detail rather than jump at single-carrier modulation and question whether this was the right choice in future.”

OFDM delivers several benefits, says Stern, especially in the flexibility it brings in managing spectrum. OFDM allows a fibre’s spectrum band to be used right up to its edge. Indeed Melamed is confident that by adopting OFDM for optical, the spectrum efficiency achieved will eventually match that of wireless.


“OFDM is very tolerant to rate adaptation.”

Amitai Melamed, ECI Telecom


The technology also lends itself to parallel processing. “Each of the sub-carriers is orthogonal and in a way independent,” says Stern. “You can use multiple small machines to process the whole traffic instead of a single engine that processes it all.” With OFDM, chromatic dispersion is also reduced because each sub-carrier is narrow in the frequency domain.

Using OFDM, the modulation scheme used per sub-carrier can vary depending on channel conditions. This delivers a flexibility absent from existing single-carrier modulation schemes such as quadrature phase-shift keying (QPSK) that is used across all the channel bandwidth at 100 Gigabit-per-second (Gbps). “With OFDM, some of the bins [sub-carriers] could be QPSK but others could be 16-QAM or even more,” says Melamed.  

The approach enables the concept of an adaptive transponder. “I don’t always need to handle fibre as a time-division multiplexed link – either you have all the capacity or nothing,” says Melamed. “We are trying to push this resource to be more tolerant to the media: We can sense the channels' and adapt the receiver to the real capacity.” Such an approach better suits the characteristics of packet traffic in general he says: “OFDM is very tolerant to rate adaptation.”

The Consortium’s goal is to deliver a 1 Terabit light path in a 175GHz channel. At present 160, 40Gbps can be crammed within the a fibre's C-band,  equating to 6.4Tbps using 25GHz channels. At 100Gbps, 80 channels - or 8Tbps - is possible using 50GHz channels. A 175GHz channel spacing at 1Tbps would result in 23Tbps overall capacity. However this figure is likely to be reduced in practice since frequency guard-bands between channels are needed. The spectrum spacings at speeds greater than 100Gbps are still being worked out as part of ITU work on "gridless" channels (see OFC announcements and market trends story).

ECI stresses that fibre capacity is only one aspect of performance, however, and that at 1Tbps the optical reach achieved is reduced compared to transmissions at 100Gbps. “It is not just about having more Gigabit-per-second-per-Hertz but how we utilize the resource,” says Melamed. “A system with an adaptive rate optimises the resource in terms of how capacity is managed.” For example if there is no need for a 1Tbps link at a certain time of the day, the system can revert to a lower speed and use the spectrum freed up for other services.  Such a concept will enable the DWDM system to be adaptive in capacity, time and reach.


Project focus

The project is split between digital and analogue, optical development work. The digital part concerns OFDM and how the signals are processed in a modular way.

The analogue work involves overcoming several challenges, says Stern. One is designing and building the optical functions needed for modulation and demodulation with the  accuracy required for OFDM. Another is achieving a compact design that fits within an optical transceiver. Dividing the 1Tbps signal into several sub-bands will require optical components to be implemented as a photonic integrated circuit (PIC). The PIC will integrate arrays of components for sub-band processing and will be needed to achieve the required cost, space and power consumption targets.

Taking part in the project are seven Israeli companies - ECI Telecom, the Israeli subsidiary of Finisar, MultiPhy, Civcom, Orckit-Corrigent, Elisra-Elbit and Optiway- as well as five Israeli universities.

Two of the companies in the Consortium

“There are three types of companies,” says Stern. “Companies at the component level – digital components like digital signal processors and analogue optical components, sub-systems such as transceivers, and system companies that have platforms and a network view of the whole concept.”

The project goal is to provide the technology enablers to build a terabit-enabled optical network. A simple prototype will be built to check the concepts and the algorithms before proceeding to the full 1Terabit proof-of-concept, says Stern. The five Israeli universities will provide a dozen research groups covering issues such as PIC design and digital signal processing algorithms.

Any intellectual property resulting from the project is owned by the company that generates it although it will be made available to any other interested Consortium partner for licensing.

Project definition work, architectures and simulation work have already started. The project will take between 3-5 years but it has a deadline after three years when the Consortium will need to demonstrate the project's achievements. “If the achievements justify continuation I believe we will get it [a funding extension],” says Stern. “But we have a lot to do to get to this milestone after three years.

Project funding for the three years is around US $25M, with the Israeli Office of the Chief Scientist (OCS) providing 50 million NIS (US $14.5M) via the Magnet programme, which ECI says is “over half” of the overall funding.


Further reading:

Ofidium to enter 100Gbps module market using OFDM

Webinar: MultiPhy on the 100G direct detect market 

Reader Comments

There are no comments for this journal entry. To create a new comment, use the form below.

PostPost a New Comment

Enter your information below to add a new comment.
Author Email (optional):
Author URL (optional):
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>