Transmission of Commercial Low Latency Interfaces Over Hollow-Core Fiber

Abstract — For the first time we successfully demonstrate the transmission of commercial 10Gb/s and 100Gb/s interfaces, both direct detect and coherent, over 2.75km of hollow-core fiber. The fiber is connected to the transceivers using spliced standard single-mode pigtails. This demonstration solidifies the commercial potential for low latency applications. Index Terms — Hollow-core fiber, photonic bandgap fiber, photonics crystal fiber, 10G, 100G I. I NTRODUCTION OLLOW - CORE FIBERS (HCF) have been proposed in the 1990’s by Philip Russell and his research group [1], covering a wide range of potential applications such as ultrafast nonlinear optics in gas filled HCF, super-continuum source generation, or particle guidance. Naturally, hollow-core fibers have also attracted the attention of telecommunication experts trying to exploit its unique properties. For one, loss in the fiber is not limited by the infrared absorption for longer wavelengths due to the absence of glass in the core. All other effects being equal, the loss in a HCF could theoretically be reduced to below 0.1 dB/km in a 2 µm transmission window [2, 3]. Second, since the light is predominantly guided in air, detrimental effects from non-linearities are reduced by orders of magnitude, enabling e.g. higher-order modulation, similar to wireless communication. Finally, the absence of glass in the core leads to a propagation speed of light that is close to the speed of light in vacuum, thus reducing latency by roughly 30%. The last property makes the fiber interesting for a range of low latency applications, with a possible focus on high frequency trading. Here, one relies on transmission technology optimized for lowest latency in order to perform high volume transactions on low margin arbitrage between various stock exchanges. Latency optimization can be done with respect to the server computing speed, geographical proximity to the stock exchange, the modem processing delay, and the transmission network. The latency objective is down to a sub-microsecond optimization. Inside data centers, high-frequency traders optimize the position of their server, trying to be as close as possible to the main switch leaving the data center. Thus, the