ALGORITHMS INTO NEXT GEN HFT.
ALGORITHMS INTO NEXT GEN HFT.
CHALLENGE
Aviat is a microwave communications networking specialist, providing dependable products, services, and support to numerous customers worldwide. In particular, Aviat specialises in high performance systems, and nowhere is this more apparent than in high frequency trading, or HFT.
HFT systems typically span vast areas. The classic route is Chicago / New York. Over these routes the equation is simple: lowest latency end-to-end wins. Microwave easily beats fibre, and so is the base technology, but it requires further optimisation.
Additionally, this market is extremely competitive. HFTs are classic first adopters and want the best, now. Companies in this space must continue to show latency innovation in the race to zero, but it becomes harder and harder to drive the latency lower and lower…
ASSIGNMENT
Aviat had an extremely capable platform, using FPGA technology to enable low latency parallel processing for wireless data transfer. The question to Qamcom was: would it be possible to continue to use this and with a software upgrade get a state-of-the-art system?
SOLUTION
The answer was: yes! But you knew that already.
The first thing we needed to do was put in place the right team. For this development we needed a team that had deep expertise in communications systems, signal processing, embedded software, and, ultimately, the physical limits of systems. We also needed people who knew how to work on products that were installed all over the world and meeting strict regulatory requirements. In New Zealand we had Dylan and Ravi, providing systems, software, and signal processing / FPGA expertise and also enabling direct contact with Aviat New Zealand. And we complemented this with a larger team in Gothenburg, Sweden.
Next, we analysed the system and determined its limits. Fortunately the FPGA-based platform was, indeed, highly capable, and we knew that we would be able to achieve our goal by developing and implementing algorithms specifically for this.
We then set to work, in parallel providing constant updates for Aviat and working with Aviat’s customers to get the system into production networks.
“A highlight for me was that we ended up with something that we believed was truly innovative and best-in-class. We came to this point because Aviat asked a simple but challenging question: ”what are the limits?”. So, we removed prior assumptions and re-looked at communication theory and how we could re-apply this using state-of-the-art algorithms plus parallel processing to get the lowest possible latency. Super cool!”
Dylan Jorgensen
Systems Engineer
OUTCOME
At the end of the day we had:
- Industry leading latency via a simple firmware upgrade to installed systems.
- A 75% reduction in latency; a major step improvement in the race to zero.
- And this was all achieved on budget and, crucially, within aggressive time frames.
“Since 2012 we’ve been involved at the cutting-edge of low latency networks. The software and firmware upgrade from Qamcom dramatically reduced the latency of our solution. It enabled us to maintain that edge.”
Matthew Hansen, VP Product Line Management, Aviat Networks
