WHY 400G FTTH#

Nicola von Thadden (PFALZKOM), Arjan Koopen (i3D.net)

• Why: A rare, fully worked optical engineering case: pushing 400G over a simplex (single-strand) fiber 50 km to Amsterdam. Directly exercises the PM's physical-networking interests (coherent optics, DWDM, loss budgets, DC/WAN interconnect) with a problem most DC folk never face — no return strand.
• What: The WHY2025 hacker festival built a temporary ISP with its own AS and needed a 400G uplink from the field to Nikhef. The path used a SURF alien-wave (one 100 GHz DWDM channel) to Alkmaar, then rented duplex KPN dark fiber plus a friendly local FTTH provider's fiber — which turned out to be simplex only. The deck documents the history of camp uplinks (HAR2009 10G → SHA2017 100G coherent → MCH2022 200G → WHY2025 400G) and how they solved the simplex problem.
• How: The core finding: current coherent optics require RX and TX on the same frequency, so a standard 400G ZR transceiver cannot be split across two channels on one strand (initial ADVA tests failed; ADVA confirmed you need different product). Two escape hatches: (1) optical circulators to share one channel — works but adds attenuation/reflection risk; (2) the "cursed idea": use two separate optics, one TX-only on channel A and one RX-only on channel B, with active DWDM/OpenZR+ Smartoptics transponders in the middle because they ignore Ethernet link state ("it's a cable"). The DWDM mux doubles as the BiDi mux. The local fiber also surprised them with ~20 dB of loss over 50 km (KPN fiber behaving like 80 km).
• Where applicable: Event-specific topology (alien waves, donated strands, temporary AS), but the technique — BiDi over simplex via dual unidirectional optics or circulators, and the coherent-same-frequency constraint — transfers to anyone constrained to single-strand fiber (FTTH plant, leased simplex pairs, some campus runs). Real measured loss figures included.