1. Introduction

A 520 km route that carries 280 km of hollow-core fiber and 240 km of standard single-mode fiber presents a planning tool with two transmission media whose noise behaviour differs by six orders of magnitude in one term, by a factor of two in another, and by the presence or absence of a third term entirely. The single-mode spans generate Kerr nonlinear interference that grows as the cube of launch power. The hollow-core spans generate almost none, because the guided field propagates in air with a nonlinear coefficient roughly 1,000 times lower than silica (measured, reported across the anti-resonant hollow-core fiber literature). In exchange, the hollow-core spans carry inter-modal interference, an impairment that does not exist on single-mode fiber at all.

Two separate budgets — one for each medium — answer the wrong question. A coherent receiver at the far end of the route reports one signal-to-noise ratio, and that number is set by the incoherent sum of every noise power generated anywhere along the path, in whichever medium generated it. The metric that carries that sum on the planning side is the generalized signal-to-noise ratio (GSNR), and the composition rule it obeys is a reciprocal sum over spans. One accounting spans both media because the physics gives no other option.

Hollow-core fiber reached the point where this question is operational rather than academic. Record attenuation stands at 0.040 dB/km at 1550 nm (measured, reported at OFC 2026), a broadband double-nested anti-resonant design measured 0.091 dB/km at 1550 nm with loss below 0.2 dB/km across a 66 THz window (measured, peer-reviewed), production fiber ships at 0.10–0.15 dB/km, and hyperscale operators have moved from pilot links to multi-thousand-kilometre procurements. The practical deployment picture is set out separately in the MapYourTech readiness assessment of hollow-core fiber. What follows is the design-side counterpart: how the noise budget is written when a route holds both media, and how the feasibility decision is made from it.

Selective deployment, not wholesale replacement, is the shape the first decade of hollow-core fiber in transport networks will take. Fiber cost remains far above single-mode fiber, so operators convert individual spans rather than whole routes, and the question a planner faces is not whether hollow-core fiber performs better but which span to convert and what the conversion buys in decibels. Answering that requires a budget that treats each span in its own physics and then adds the results correctly.

The scope of this article is the physical-layer noise budget on optically transparent mixed-media routes using coherent transponders, from the definition of the metric through the per-span ledger, the launch-power decision on each medium, the end-to-end feasibility test, and the commissioning and monitoring steps a mixed route needs. Regeneration, restoration routing and the economics of fiber procurement bound the argument and are otherwise outside it. Every worked value carries forward from one section to the next, so the 520 km reference route defined in Section 4 is the route whose margin is tested in Section 7 and whose span ranking is derived in Section 8.

2. Generalized Signal-to-Noise Ratio Definition and Component Terms

Generalized signal-to-noise ratio is the ratio of per-channel signal power at the receiver to the total power of every additive noise source accumulated along the path, expressed in decibels referred to a 0.1 nm reference bandwidth. It differs from optical signal-to-noise ratio in counting nonlinear interference and inter-modal interference alongside amplified spontaneous emission, which makes it the quantity that a coherent receiver's reported signal-to-noise ratio tracks.