Over the last few years, a new breed of specialised service providers have been offering low latency wireless Point-to-Point networks between financial centers. But for the High Frequency Trading (HFT) firms who use these services, fast is not enough. To beat the competition, they want their connection to be faster than everyone else.
Secretive cash rich financial trading firms are already renting space on towers and are probably building their own wireless networks using millimetre waves (the same frequencies the mobile industry plans to use for future 5G networks). Considering the huge sums at stake for the winners of this race, it wouldn’t surprise me if they have already built in-house technologies while the mainstream mobile industry has only reached the planning stage.
Low latency is absolutely critical for HFT firms. The financial firm that can connect to the marketplace before its competitors stand to make billions in profits. For these fintech players, the fiber backbone is just not fast enough. For example, the lowest latency (delay) between London and Frankfurt in fiber is 8.35 milliseconds while the speed of light in free air is only 2.1 ms. Specialised wireless fintech service providers such as Perseus and McKay Brothers have managed to get the latency down to around 4.6 ms on this route. Considering that the theoretical floor for latency is 2.1 ms there is plenty of room for aggressive HFT firms to build their own optimised network and get below 4.6 ms.
Fiber’s “slowness” is due to the fact that cables don’t run in a straight line of sight between two cities. Another reason is that the speed of light in fiber is 33 percent slower compared to the speed of light in free air. When the signal traverses through the network, latency is also added each time it passes through a router.
Financial players work hard to reduce latency in every part of their infrastructure. A seemingly minor difference such as the location of computers on different floors in a building can add to latency. In one example, the latency on the 2nd floor was 0.184 ms, but on the 9th floor it was 0.183 ms.
Building a Point-to-Point network (with microwave links transmitting narrow beams between a line of high towers) is an old technology that was used for long range communication before fiber optics. This recent revival of wireless has been made possible by better RF components (in the high gigahertz bands) and ultra-fast chips that can handle the signal processing without adding much latency.
This type of backbone network will never be cheaper than using fiber. The need for free line of sight and the curvature of earth puts a limit on the longest distance between towers. It is possible to increase reach by building higher towers but increased height adds to construction costs. Wind drift of the towers and path loss due to rain attenuation are other problems that have to be overcome. Transmission capacity can be very high if wide enough carriers are used in the (idle) millimetre wave bands above 30 GHz, though this technology will always be dwarfed by fiber.
We don’t know exactly how far the financial HFT firms’ secret in-house projects have come. But one thing’s for sure – they are not being held back by slow moving industry committees. They are most likely using GHz/millimeter waves but another solution could be lasers (from AOptix?). As the main objective is to reduce latency, my guess is that they’re also experimenting with transmission of some form of low level “raw” signal where the IP headers of the packets have been stripped away.
Even though this highly specialised fintech transmission network can be viewed as a custom built race car, it is relevant for the wider mobile industry. For example, a crucial building block for the future 5G mobile is wireless backhaul in the millimeter wave bands (+30 GHz). One of the goals in the 5G mobile specification is a sharp reduction in latency. These are areas where the fintech players appear to be years ahead. Their solutions have already been deployed, or will be in the near future.
If these secretive players are ever willing to share their technologies, they could serve as important proof of concept installations for the rest of the tech sector. And if their solutions are ever licenced, new advanced technologies may enter the mobile market from an unexpected industry – fintech. It would certainly be prudent for the mainstream mobile market to pay attention to innovations in this field.