Science AMA Series: I research new hardware and technology for internet communications and optical computers, I'm Daniel Blumenthal, AMA!

Abstract

Hi! I'm Daniel Blumenthal, a Professor of Electrical and Computer Engineering specializing in optical communications and photonic integration at the University of California – Santa Barbara, Fellow of the National Academy of Inventors, The Optical Society (OSA) and of the IEEE.

Today’s high-speed optical communications technologies work to accelerate the way we work, live and play by connecting people and computers around the world over high speed communication pipes. My lab develops new hardware and communications technologies to solve complex communications, transmission, switching and signal processing problems out of reach with today’s technologies. The primary undertaking of our research is to develop new functions integrated on small chips called photonic circuits, and use these circuits to build networks in ways that save energy and increase the scale of connectivity and bandwidth of data centers and the Internet. We are, in theory, greening future networks while allowing them to scale to accommodate future applications and the systems that rely on those networks.

I am a Professor in the Department of Electrical and Computer Engineering at the University of California, Santa Barbara and Director of the Terabit Optical Ethernet Center (TOEC) and head of the Optical Communications and Photonic Integration (OCPI) group (ocpi.ucsb.edu). I have served as PI for large-scale research programs including DARPA/MTO funded CSWDM, LASOR and iPhod projects at UCSB. In addition, I have served on the Board of Directors for National LambdaRail (NLR) and Internet2 Architecture Advisory Council. I have co-founded two companies, Packet Photonics, Inc and Calient Networks. My research interests are in optical communications, photonic packet switching and all-optical networks, ultra-low loss optical waveguides and silicon nitride photonics circuits, all-optical wavelength conversion and regeneration, ultra-fast communications, InP Photonic Integrated Circuits (PICS) and nanophotonic device technologies. I have published over 410 journal and conference papers, 5 book chapters, and co-authored a leading book on tunable lasers. I have served on technical program committees of international conferences including the conference on Optical Fiber Communications (OFC) and as guest editor for multiple journal special issues.

I will be back at 1 pm ET to answer your questions, and I look forward to sharing with you today. Feel free to Ask Me Anything!

Hi, Prof. Blumenthal. Thank you for doing this AMA.

I've read about Moore's Law(An observation that the density of transistors on a chip(and thus the computing power) doubles every two years), and was wondering if data transfer has any such equivalent?

useful_person

We see data transfer at many levels, there is on chip, between chips, board to board, intra rack, short haul, medium and long haul. These distances and their bandwidth requirement play a key role in your question. What we are seeing it that once we get off the chip, the ability to use optics to transmit data at a certain bit rate over a certain distance can exceed Moore's law. But people solve that problem by parallelizing heavily, so electronics keeps keeping up. We dont have the equivalent today for Moore's law in photonics, just curves that show fiber transmission bandwidth exceeds Moore's law. As photonic chips become denser with more components, we will probably start to get a handle on a Moore type law for photonics. But I imagine this will take 10-20 years.


Welcome,

Where and/or what are the bottlenecks in existing communication hardware?

adenovato

Today it is in the chip level I/O. A lot of information can be processed on an electronic chip, state of the art today is 12.5Tbs on a single layer 2 routing chip used in data center top of the rack switches. These chips have 40 inputs and outputs running at 100Gpbs, 200Gbps and soon 400Gbps per port, which is amazing! But the problem now is interfacing all those ports to optics. The board level routing, and density of optical hardware, and all the communications including clock and data recovery, it just keeps adding up and at these capacities is starting to become and engineering and systems build out issue. So today I think that is one of the key bottlenecks people are looking at.


Why are photonics circuits faster and greener than our current technologies? What are the biggest technical hurdles in their development and adoption?

Thanks for your time!!

kiri-kin-tha

I think it is more the opportunity for photonic circuits to be greener. This will come, if we can make it happen, from lower the power consumption of large scale computation systems, data centers and possibly the internet of things (think sensors). There are certain measurements, most use the analog nature of light interacting with material, to make a measurement or perform a function. This is sometimes more energy efficient to do with a dedicated photonic circuit (like an interferometer on a chip) than the digital electronic equivalent. Another area for example is digital fiber transmission. As the fiber capacity keeps scaling, DSPs are increasingly relied on to undo a lot of the degradation induced by transmitting data on high capacity links. These DSPs are costly and can burn a lot of power. We are back to looking at ways to put ultra low loss optical processing elements in the fiber transmission link, to offload functions from DSPs, and in some cases possibly eliminate them, to greatly reduce the power consumption of a link. This makes the link both greener, and addresses the issue of how much I/O bandwidth can be packed into a line card used for a transmission link. Biggest hurdles are getting performance to where it needs to be which is getting there, then cost by making any solution compatible with wafer scale foundry processes and lastly the packaging is a huge one in the end.


How/when did you know did you get into this field? What sort of pathway would someone need to take to get to your sort of position?

therockethornet

It was a long road for me! I did some state of the art research in ultra fast optics and optically generated microwave pulses as an undergrad in Gerard Mourou's group at the Laboratory for Lasor Energetics, then went to industry at StorageTek to work on optical data storage (before the CD came out). While in industry I started to get the feeling that I wanted to be in academia because I was very interested in not only the research and development but also continual learning and communication between the different product groups at the technical level. Then went back to get a Master's at Columbia and that is when I really started in telecommunications and data communications using photonics and integrated optics in the NSF Center for Telecommunications. Much of my background from undergraduate research in ultrafast lasers and such were applicable and I started to learn about optical CDMA and that is where the work on optical switching and optical packet switching started. I then went to work for one of the pioneers of the internet, David Farber, at UPenn, during the days when TCP/IP was being developed and it influenced my optics device and systems work to incorporate the many layers of communications protocols for the internet. Following my PHD at UC Boulder, where I focused more on self-routing optical packet switching, it wasn't until a year before writing my dissertation that I realized I wanted to be in academia for sure, but once I realized I became focused on just that, finding a teaching and research position. At that point, going first to Georgia Tech and then UCSB, I knew that I wanted to lead the path on a new type of research where the group did work in both systems and devices and underlying technologies. This was not the norm at the time, people usually focused on just one area. So it meant we truly had to be experts in both, and I was driven by doing both so we could bridge the communication gap between devices and systems/network communities. It took many years, but we did it by getting into the cleanroom making devices and being a core part of that community as well as our work building systems in the laboratory. So I guess this is just an example of what I did. It takes believing in what you are doing, having persistence, and working with good people as well as trying to see what people are doing in the present and thinking about what is next to help define the future fields. Hope that helps a little!


Might be a big ask, but can you give a laymen explanation on what exactly a photonic circuit is?

sciencereader3455

HI thank your for a great question! I'll try to keep it brief :-) A photonics circuit is a collection of optical wires, called waveguides, and other elements all put on a single chip to form a circuit or a function much like what an electronic circuit with transistors does. Just like electronic the wires are use to connect and route signals, in this case photons. Now on a photonic circuit we also have electrical connections to light emitter, like lasers, detectors to receive the light, modulators to put information on the light, etc. So a photonic circuit really means using photonic but interacting with the atoms and electronics, hence the name photonics. Other elements on this circuit can be filters, gain blocks, phase modulators and the photon analogy of many electronic components. We dont have the exact match of a resistor, capacitor and inductor and transistor so to speak but we try to mimic those functions with optical counter parts. So in the end we layout all these elements using masks, and fabrication techniques to make a chip that mostly handles information in the form of light, but also interacting with electronics.


What are the best future applications of your work?

sciencereader3455

That is a big one. We have a huge amount of bandwidth we can cover from the UV, visible to near IR and far IR. So a list of applications could include communications, spectroscopy, atomic clocks, position and navigation, biological tissue analysis, high performance analog and digital computing, optical coherent tomography, microwave photonics, remote sensing (like Lidar), quantum communications and computation, bio-sensing, frequency synthesis and a whole lot more!


Will your work make my internet faster... some day?

  • What are the primary benefits of optical circuitry over electron gating?

  • Will optical comms be more widespread in the future?

  • Can optical communication advances fix some of the roadblocks we experience today in communication networks?

  • What's beyond optical communication?

scienceaccount103040

Will try some short answers to your great questions! * really the faster you modulate data or more information per spectral bandwidth and this sent over longer distance is were optics wins out. This has been typically over 100 meters or more but we are now starting to see chip to chip and possibly on chip communications being a win. So it first is about optical wires and the lasers and detectors on the ends of those wires that can move data with less power as the speeds go up. Optical circuitry is more complicated. Lets take the easiest one. Optical switches can move data between ports, with much more bandwidth, more energy efficient than their electronic counter part. There is a caveat to this, if you need to do something to the data while switching, it is still hard to beat electronics.

  • We are seeing optical comms make their way from long haul links, to all the links inside a data center (100m) and now inside a rack, (3m) and I believe we will see optical backplanes and boards as well as on chip optics so yes more widespread.

  • Yes, I think this question is very related to the above. It all comes down to power consumption, cost, and capacity.

  • Not sure what you mean by beyond optics, but if I were to guess it would action at a distance, that might still involve photons I would guess. But who knows, communications with elementary particles? Anything is possible, that is the excitement of the future and new generation of scientists and engineers are needed to make this happen!


Will your work make my internet faster... some day?

  • What are the primary benefits of optical circuitry over electron gating?

  • Will optical comms be more widespread in the future?

  • Can optical communication advances fix some of the roadblocks we experience today in communication networks?

  • What's beyond optical communication?

scienceaccount103040

Oh, faster internet, yes the backbone, but to you as a user, that will depend on your service provider and FCC rules, in the end :-)


Oh, and what's the applicability of nanophotonic tech?

scienceaccount103040

Mostly about getting the feature size of photonic and optical components down below wavelength size so we can make densely integrated circuits. We used to think that photonics circuits would be big due to wavelength of light but we know that is not true now. Nanophotonics also lets us use new physical interactions between light and matter to do new engineering and devices that open up new functions and applications.


Additional Assets

License

This article and its reviews are distributed under the terms of the Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and redistribution in any medium, provided that the original author and source are credited.