Optical power monitoring: the leading indicator FTTH ISPs still under-use
Optical Tx/Rx power is the single most valuable telemetry an FTTH ISP has. It is also the least-used. A study of why, and what the operational payoff looks like.
Every ONU on the planet reports its optical transmit and receive power to the OLT. Every operator has the data. Almost no operator uses it as a leading indicator. This is a short study of why that is, and what changes when it is instrumented properly.
The underlying specifications are well-established: [ITU-T G.988](https://www.itu.int/rec/T-REC-G.988/en) defines the OMCI management information base, including the ANI-G (ANI-Generic) managed entity that exposes optical parameters. Every compliant ONU exposes Tx/Rx power in dBm through OMCI, which the OLT then makes available upward through SNMP, gNMI or the vendor's NMS API.
Why the data is under-used
Three reasons: (1) most NMS tools poll at 5- or 15-minute intervals, which is far coarser than optical drift needs; (2) very few tools retain per-ONU historical baselines, so operators have no reference against which "1 dB degradation" means anything; (3) the alarm rules are usually generic thresholds, not per-unit baseline offsets.
What "instrumented properly" means
- 1Per-ONU sampling every 60 seconds during business hours, at 5 minutes off-hours.
- 2Per-ONU installation baseline captured at ONT authorisation time; re-baseline on a physical work-order that touches the fibre.
- 3Alarm rule: 2 dB drift from baseline, sustained across 5 samples. Not "Rx power below −25 dBm" in the abstract.
- 4Correlate with weather + temperature at the OLT site; humidity and freeze-thaw drive most seasonal drift.
The operational payoff
Splices degrade slowly. A typical outdoor splice failure looks like: month 1 baseline, month 4 begin drift, month 6 threshold cross, month 7 hard failure. Two-to-three months of warning is the norm, not the exception. Netxol NMM captures per-ONU history by default; see [Netxol NMM](/products/modules/nmm) and the [OLT management solution](/solutions/olt-management).
2–3 months
Typical lead time on optical splice failures
Assuming baseline capture + threshold alarm.
~80%
Of splice-caused outages become preventable
When properly instrumented.
40–60%
Reduction in unplanned FTTH outage minutes
In one 3-POP deployment we observed.
A note on vendor differences
OMCI exposes the primitive; how OLT vendors surface it upward varies. Huawei via MA5800 exposes it well through the U2000 API; ZTE C620 has slightly different attribute paths; FiberHome depends on OLT firmware version. A vendor-neutral abstraction (Netxol's or otherwise) collapses this — for the vendor-abstraction detail see the [multi-vendor OLT / GPON scaling study](/blog/multi-vendor-gpon-scaling-study).
If you do one thing this quarter
Enable per-ONU optical Rx power capture at 1-minute intervals across the entire estate. Do not build the alarm rules yet — just capture the baseline. Six months of history will make the alarm rules obvious.
Standards + references
- ITU-T G.988 (OMCI base)Defines the ANI-G optical parameters.
- ITU-T G.984 (GPON base)GPON reference.
- Netxol NMMPer-ONU history + baseline threshold rules.
- Netxol OLT management solutionThe vendor-neutral abstraction.
- Related: Multi-vendor GPON scaling studyWhat breaks at 20K → 200K ONTs.
- Related: TR-369 (USP) adoption 2026The management-plane story alongside.
