John Papandriopoulos

I am currently with SnappyLabs, a San Francisco based startup.

My last role was with Adaptive Spectrum and Signal Alignment (ASSIA), Inc., a San Francisco bay area startup company. ASSIA develops systems and technology to enable the highest speeds on the world's DSL lines.

I grew up in Melbourne and absolutely love Australia. Everyone calls me jpap, and so can you.

Whilst still in Australia, I was a Research Fellow with the ARC Special Research Centre for Ultra-Broadband Information Networks (CUBIN), in the EEE department at the University of Melbourne. I completed the Ph.D. degree in 2006 at the same institution, for which I was awarded the Chancellor's Prize.

Further back still, my undergraduate studies were completed at RMIT University, where I received B.E. (Communication) and B.App.Sci (Computer Science) degrees in December of 2001, along with the J. N. McNicol award (University Medal).



John Papandriopoulos
Wed 25 Oct, 2006 (with Ph.D. submission balloon)

As a Victoria Fellow I was propelled to some wonderful institutions in the USA in late 2005. In 2006 I was fortunate to visit France as part of an AFAS FEAST-France Fellowship, in addition to some other top telecommunications research labs in Europe.

Outside of my research efforts, I've spent two years as the UniMelb IEEE Student Branch chairperson, and in parallel, as the branch webmaster. I helped create the CUBINlab research network in August of 2002 and was the core-network system admin for several years.

FAQ on my DSL Research

Following much media coverage surrounding my DSL-related Ph.D research, including a great plug on Slashdot I've received a flood of questions from the world over. Below you can find some answers to those most frequently asked.

  1. What speeds can we expect from your technology?

    Digital Subscriber Line (DSL) technology allows fast broadband network access over regular copper telephone wiring.

    Various flavors of DSL technology offers maximum theoretical aggregate (i.e. downstream + upstream) speeds of 9Mbps (ADSL), 25Mbps (ADSL2+), and 250Mbps (VDSL2), though these maximum speeds are not always attainable in practice. Many DSL subscribers in Australia, for example, are limited to speeds around 1-2Mbps (ADSL); less than a quarter have access to higher ADSL2+ speeds of around 20Mbps.

    Regulatory constraints aside, two main factors limit DSL speed: long line lengths and crosstalk interference. Crosstalk interference is caused by electromagnetic coupling between telephone wire pairs in ducts on their way to the exchange---a result of the relatively poor quality of wiring used for traditional telephony. Crosstalk can cause major problems even on shorter lines, and depends on the network topology. This interference is conceptually similar to the faint conversations of a crossed line once problematic decades ago. With the push for higher bandwidths, this crosstalk problem has returned for data-transmission over DSL and solving it is a crucial step in enabling fast broadband speeds with DSL.

    Our technology, developed as part of my Ph.D. thesis work with advisor A/Prof. Jamie Evans, aims to manage this crosstalk interference, consequently allowing telecommunication providers to maximize the data-rates of their networks. We can do this dynamically, and adaptively, to try and get the "best compromise" of interference between neighboring lines to maximize performance. In research circles, this is known as "Dynamic Spectrum Management" (DSM).

  2. When will your technology be available?

    With the assistance of Melbourne Ventures, the technology commercialization Company for the University of Melbourne, we have filed for two patents relating to these technologies.

    The University is seeking parties who are interested in licensing the technologies, so that it can be made available to you in the future.

    This process will naturally take some time, and it could be several years before it is implemented by equipment vendors and then installed in telecommunication networks and enjoyed by end-users.

    Please contact , Melbourne Ventures Pty. Ltd., on +61 3 8344-3066 (voice) for further information.

  3. How does your technology work?

    We formulate a (standard) model for how DSLs attain their data-rate, considering crosstalk interference effects. We exploit the fact that interference varies across the frequency spectrum, and that if we are clever about the way in which we allocate transmission power amongst the various frequency bands, we can manage the interference between (strongly) coupled lines. For example, one user that interferes strongly with others within a band may be "turned off" within that band to avoid damaging the other lines, consequently raising the overall network rate.

    The model is embedded into a mathematical optimization problem that aims to maximize the data-rate of the network (other variations also exist) systematically. It results in a spectrum for each user that is shaped appropriately to minimize interference and maximize data-rate.

    Our research has culminated in two broad methods, called SCALE and SCAPE, that attempt to solve this (inherently difficult) mathematical optimization problem. We do this with low computational complexity by breaking the problem down into a more simple form that can be solved easily; then adapting the simplified model according to the solution and repeating iteratively.

    Several theoretical results have been established relating to these methods; see my research page for further detail.

  4. Do I need to buy a new modem?

    It depends. Our research may be applied to current-day ADSL2+ and VDSL networks to bring speeds closer to the theoretical maximum without any major changes to end-user modem equipment. For the full benefit of our technology, more substantial changes may be required. This, in turn, may require the purchase of a new modem; in some cases, equipment manufacturers may be able to implement a firmware update to take advantage of our enhancements. Don't generally expect, however, that a modem designed for a particular DSL technology (e.g. ADSL2+) can be magically upgraded to one of a later generation (e.g. VDSL2) with software magic.

  5. Is it really practical or just a theoretical academic exercise?

    There is a hive of research activity in the area of dynamic spectrum management (DSM). Much of the research activity has produced methods and algorithms that require excessive computational resources that are difficult to implement.

    Our research surrounding SCALE and SCAPE are practical methods that exhibit very low complexity that can be implemented in practice.

    I recently joined the team at startup company ASSIA, where I am contributing to their efforts in making the DSM dream a reality.

  6. Are there any limits to broadband speeds over copper telephone lines?

    Yes, of course! You might be surprised, however, at just how high these limits can be. Newer DSL technology can already support total aggregate rates of up to 250Mbps (VDSL2).

    Research by Prof. John Cioffi's group at Stanford University has recently concentrated on the next-wave of DSL technology, that can offer speeds of 1-2Gbps over copper. That research, however, is likely to be many more years away from implementation and deployment. Our technology can assist with raising broadband speeds over DSL in the much shorter term.

  7. Can I have a copy of your thesis?

    My Ph.D. dissertation is currently bound by a confidentiality agreement due to intellectual property issues. It will be made available on my website in PDF form as soon as these restrictions are lifted.

  8. Are you available for a media interview?

    Of course! Please contact , Media Officer at the University of Melbourne, on +61 3 8344-7220 (voice).