Optical components

Axonlink: A New Take on Tunability

What could turn out to be a breakthrough technology for optical receivers was unveiled yesterday by Axonlink Inc. (see Axonlink Intros AWDM).

The technology, called optical heterodyning, promises to eliminate the need for expensive filters and multiplexers in access networks, by allowing the receiver to pick out one channel in the presence of other channels. It's not a new technology -- in fact, it works in the exact same way as ordinary radio receivers -- but this appears to be the first time it's been commercialized for optical applications.

This is the first time Axonlink has talked about its plans since it was spun out of Israel's ECI Telecom Ltd. (Nasdaq/NM: ECIL) in late 2000. Axonlink was founded by Joshua Piazetsky and Hertzel Yehezkely, who are the CTO and deputy CTO, respectively, at ECI. Yehezkely served as Axonlink's CEO until just this week, when he takes on the positions of president and COO, handing over the CEO reins to newcomer Dror Nahumi.

According to Nahumi, there was plenty of research into optical heterodyning going on in the 1980s. "The motivation at the time was completely different to today," he says. "One of the properties is increased receiver sensitivity, because you are also doing amplification of the incoming signal." Back then, Erbium Doped-Fiber Amplifiers (EDFAs) hadn't been invented, and researchers were using the extra receiver sensitivity to make optical signals go as far as possible. When EDFAs were invented in 1987, most of the work in the field stopped.

Today, the main motivation for developing the technology is to achieve maximum performance at minimum cost. Increased receiver sensitivity allows cheaper types of detectors to be used. Tunability in the receiver, as noted, eliminates the need for some other components in the network. And it also makes the receiver a perfect companion for a tunable laser.

The best way to understand how optical heterodyning works is with the aid of this diagram, supplied by Axonlink:

Heterodyning works by mixing the incoming signal -- in this case a bunch of DWDM channels carrying data -- with another signal of similar, but not identical, wavelength. This is implemented in the receiver by combining light from the input fiber with light from an on-board laser (the combiner is shown as a circle).

All of the light then falls on a standard PIN photodetector, which converts it into an electronic signal. This information can be filtered electronically to extract the data from any one of the incoming DWDM channels.

The filtering relies on a process called "beating", which makes the amplitude of the electronic signal wobble at a certain frequency -- called the intermediate frequency (IF) -- that is always equal to the difference in frequency (wavelength) between the two incoming optical signals. To pick out one channel and ignore the others, the receiver simply needs to know where to look -- in other words it just needs to know the intermediate frequency.

In practice, the intermediate frequency is fixed, and the receiver tunes by shifting the wavelength of the onboard laser. Axonlink does this in the current version of its device by heating and cooling standard Distributed Feedback (DFB) Lasers, creating a tuning range of a few nanometers.

One way of cutting costs is to split the output of the on-board laser and use some of it in the receiver and the rest of it to power a transmitter. Indeed, that is what's shown in the diagram.

Sharing the laser in this way does mean that the transceiver will send and receive signals at different wavelengths. But that's okay, says Nahumi. "For applications where you want to have bidirectional capabilities -- sending signals in both directions on the same fiber -- you like to have channels at different wavelengths." This type of module is likely to find applications in access and fiber-to-the-home scenarios. In other applications, he adds, where it's important to send and receive at the same wavelength, two lasers can be used inside the module.

Nahumi says the company recently demonstrated a tunable transceiver like the one in the diagram to potential customers. Now it is exploring the business case for different applications, while also looking for a partnership with a larger vendor.

Axonlink has gotten first-round funding of $12 million from Benchmark Capital, ECI Telecom, and Innovacom.

— Pauline Rigby, Senior Editor, Light Reading
pavlovsdog 12/5/2012 | 12:42:33 AM
re: Axonlink: A New Take on Tunability Sure you can heterodyne two lasers, but that requires aligned polarizations. Where is polarization control (and cost) in the diagram?
lightbridge 12/5/2012 | 12:42:32 AM
re: Axonlink: A New Take on Tunability Optical heterodyning requires the polarizations of the signal and the local laser to coincide. That can be achieved by polarization tracking (polarization controller with electrical feedback) or by using a polarization diversity scheme (local laser in 2 orthogonal polarizations, 2 photodiodes, electronic processing). Both add a lot complexity to the receiver.
Anybody knows, what approach Axonlink has taken?
isitso? 12/5/2012 | 12:42:31 AM
re: Axonlink: A New Take on Tunability There another way todeal with polarization issue - you simply scramble the poalrization by modualting teh polarization of the incoming light
with twice the signal bandwidth. Ine modulator can do it for all the channels
This is a simple method, albeit it comes with a few dB penalty
LightBeating 12/5/2012 | 12:42:08 AM
re: Axonlink: A New Take on Tunability isitso,

That means you also need that depolarizer with your own transmitter. So the Rx/Tx module must contain one cooled DFB laser, one photodetector, the depolarizer, and all the demodulation and tracking electronics. Sounds like a pretty expensive device to me.

The need for polarization diversity is probably what killed optical coherent communications schemes, more than the advent of the EDFA.

On the other hand, heterodyning is a powerful demultiplexing technique, and you can probably pack many more channels within the same bandwidth than by using passive mux/demux. A DFB laser can be tuned over a couple of nm easily, and that's 250 GHz. Say you get a spectral efficiency of 0.6 (as opposed to 0.2 or 0.4 with passive components), that could mean 150 Gb/s, more than enough for LAN's, WAN's or metro networks, with the added flexibility of tunable transmitters and receivers. Of course, you probably couldn't use standard DFB's because of the chirp, which eats up too much bandwidth, so a DFB+EA modulator would be required ($$$), and you're still stuck with that damned polarization diversity issue. So cost-wise, we're not there yet, compared with run-of-the-mill CWDM, for example.

lightbridge 12/5/2012 | 12:41:59 AM
re: Axonlink: A New Take on Tunability LB,

Heterodyning doesnGÇÖt really improve the spectral efficiency. As it is pretty difficult to design an optical phase locked loop (PLL), the IF needs to be about 1.7 times the bit rate (B). If the required receiver bandwidth is 0.7*B, the IF filter needs to pass frequencies between B and 2.4*B. To avoid the beat between the next neighbor channel and the local laser falling into the IF filter band, the neighbor channel must be separated by at least 5 times the bit rate. This yields a spectral efficiency of 0.2, the same as for DWDM.

Sign In