Ch 13 — Optical Test, Measurement, and Field Commissioning

Optical Test, Measurement, and Field Commissioning

Field commissioning and fault isolation in an optical transport network use a small but specific toolkit: the OTDR for fibre-plant characterisation and event location, the OSA for wavelength-domain power and OSNR measurement, the BERT for client-signal layer verification, coherent-DSP-derived telemetry for in-service line-side metrics, and TCM (Tandem Connection Monitoring) for segment-by-segment fault isolation across multi-operator OTN paths. This chapter covers each tool’s operating principle, its limits, the failure modes it identifies, and how they combine in a structured turn-up checklist that takes a span from fibre-laid through to client traffic.

Concept	What it says
OTDR	Optical Time-Domain Reflectometer. Sends a pulse down the fibre and timestamps the backscattered light to map distance vs return loss. Locates events (splices, connectors, breaks), measures span loss and splice loss, and characterises uniform fibre attenuation.
OSA	Optical Spectrum Analyser. Sweeps a tunable filter across wavelength to measure per-channel power and out-of-band noise. The standard tool for OSNR, channel-power flatness, and spurious-emission verification.
In-band OSNR	With flex-grid and dense channel spacing, traditional out-of-band OSNR measurement is unreliable because there is no noise-only spectral region between channels. In-band techniques (polarisation-extinction, coherent-DSP-derived) measure noise underneath the signal.
BERT and FEC margin	A Bit Error Rate Tester verifies the digital client signal against a known pattern; FEC pre-correction BER and Q-margin computed by the coherent transponder give the line-side equivalent in-service.
TCM — Tandem Connection Monitoring	ITU-T G.709 mechanism that lets a sub-segment of an OTN connection be monitored independently of the end-to-end path. Decisive for inter-operator fault localisation: “the carrier’s segment is bad” rather than “the whole circuit is bad”.

OTDR — Optical Time-Domain Reflectometer

An OTDR launches a short, high-power pulse into the fibre and records the back-reflected light over time. Two physical mechanisms produce returns:

• Rayleigh backscatter — uniform, gradual return from microscopic refractive-index inhomogeneity in the fibre. The slope of the backscatter trace gives fibre attenuation in dB/km.
• Fresnel reflection — discrete, large returns from refractive-index discontinuities (connectors, mechanical splices, the fibre end). Magnitude is reported as reflectance in dB.

Time-of-flight gives distance via the fibre’s group index of refraction (typically n_g ≈ 1.4682 for SMF at 1550 nm), so a return at 1 µs round-trip corresponds to ~102 m one-way.

Reading an OTDR Trace

Trace feature	Physical cause	Measurement
Linear downward slope	Rayleigh backscatter from fibre	Attenuation in dB/km — typically 0.20–0.22 at 1550 nm for G.652.D
Sudden vertical drop with no spike	Fusion splice	Splice loss in dB (read as the trace step)
Sudden drop with reflective spike	Connector or mechanical splice	Loss in dB + reflectance in dB
Long flat region followed by noise floor	Fibre end	End reflectance
Tall reflective spike, no loss	Open connector (Fresnel reflection from glass-air interface, ~–14 dB)	Diagnostic: connector unmated
”Ghost” event	Multiple internal reflections in earlier event	Identifiable by distance = 2× original event

Dead Zones and Launch Cable

The OTDR’s near-field is unusable because the launch pulse dominates the photodiode return. Two dead zones matter:

• Event dead zone — the minimum distance at which two events can be distinguished. Typically 0.5–5 m for short-pulse modes.
• Attenuation dead zone — the minimum distance from a reflective event before the trace’s logarithmic slope can again be read. Typically 5–30 m.

Rule of Thumb

Always use a launch cable (typically 100–500 m of patchcord coiled at the OTDR end) when characterising a span. The launch cable hides the OTDR’s dead zone in a “throwaway” stretch, so the first connector / splice on the actual fibre under test is visible. Symmetrically, a receive cable at the far end lets the operator verify the far-end connector loss; without it, the fibre-end reflection masks the last connector’s contribution.

flowchart LR
    OTDR["OTDR<br/>instrument"] -->|"connector"| LAUNCH["Launch cable<br/>200 m coil"]
    LAUNCH -->|"connector under test"| FIBER["Span under test<br/>~80 km G.652.D"]
    FIBER -->|"connector under test"| RECV["Receive cable<br/>200 m coil"]
    RECV --> END["Far-end terminator<br/>or APC connector"]
    style OTDR fill:#378ADD,stroke:#185FA5,color:#fff
    style LAUNCH fill:#7F77DD,stroke:#534AB7,color:#fff
    style FIBER fill:#1D9E75,stroke:#0F6E56,color:#fff
    style RECV fill:#7F77DD,stroke:#534AB7,color:#fff
    style END fill:#BA7517,stroke:#854F0B,color:#fff

Warning

An OTDR measures two-way return but reports one-way distance and loss. The instrument knows it is reading round-trip backscatter and divides the loss by two automatically. This works correctly only if the fibre’s backscatter coefficient is uniform — at a fibre splice between two different fibre types (e.g. G.652.D ↔ G.654.C), the splice can appear as a gain event from one direction and a loss from the other. Always perform bidirectional OTDR (measurement from both ends, average the two) on mixed-fibre spans.

OSA — Optical Spectrum Analyser

An OSA sweeps a tunable optical filter across wavelength and reports the power passing through the filter at each tuning. Resolution bandwidth (RBW) is typically 0.05–0.1 nm — corresponding to ~6–12 GHz at 1550 nm. The output is the optical-domain analogue of an RF spectrum analyser: a power-vs-wavelength plot.

Out-of-Band OSNR Measurement and Its Limits

The classical OSA-based OSNR measurement (IEC 61280-2-9) is straightforward:

1. Measure peak channel power P_signal in dBm at the channel centre.
2. Measure noise floor power P_noise between channels (or at a guard-band region) at the same RBW.
3. Compute OSNR = P_signal − P_noise + 10·log10(RBW_meas / 0.1 nm).

The 10·log10 correction normalises to the 0.1 nm reference RBW.

This works perfectly for traditional 100 GHz-spaced 10 G channels, where the 100 GHz spacing leaves abundant noise-only spectral region between signals. It breaks down with flex-grid and dense modern formats:

Channel scenario	Noise-only region between channels	Out-of-band OSA OSNR?
100 GHz spacing, 10 Gbit/s NRZ	~80 GHz clean	Reliable
50 GHz spacing, 100 Gbit/s PDM-QPSK (32 Gbaud)	~10–15 GHz clean	Marginal — measures NLI partly as noise
75 GHz flex-slot, 400 Gbit/s 16-QAM (64 Gbaud)	~5 GHz clean	Unreliable — bleed-through dominates
50 GHz flex-slot, 800 Gbit/s 64-QAM (96 Gbaud)	None	Impossible — channel occupies full slot

Warning

Quoting OSA-based OSNR for any flex-grid coherent service is misleading at best and wrong at worst. The OSA measures the gap power, but the gap is full of nonlinear-interference (NLI) noise plus filter-skirt bleed from the adjacent channels. The result is typically several dB more pessimistic than the actual signal-quality margin the receiver experiences.

In-Band OSNR Methods

Two production techniques measure noise underneath the signal:

Method	How it works	Accuracy	When to use
Polarisation-extinction (P-OSNR)	Signal is polarised; ASE noise is unpolarised. Tunable polarising filter aligned to nullify the signal exposes the noise alone	±0.5 dB	Field-portable test sets, third-party verification
Coherent-DSP-derived OSNR	Modern coherent transponders compute OSNR internally from the received constellation’s noise variance, after equalisation removes deterministic distortion	±0.3 dB	Always available in-service on coherent links — the operator’s primary OSNR readout
Modulated/unmodulated swap	Compare powers with TX modulator on vs off	±1 dB	Lab/commissioning only — disrupts service

The coherent-DSP-derived OSNR has fundamentally changed the operations workflow: instead of dispatching a technician with a portable OSA, the NMS or controller streams real-time OSNR per circuit via gNMI telemetry (see 12-optical-control-plane-and-automation).

BERT — Bit Error Rate Tester

A BERT generates a known pseudo-random or pre-defined pattern (PRBS31, PRBS23, K28.5 idle, fixed user pattern) at a specified line rate and counts errored bits at a far-end receiver. For OTN turn-up, the standard sequence is:

1. Loopback at far-end — internal facility loopback at the far transponder, BERT at near-end transmitter and receiver.
2. PRBS31 line pattern, ≥10 minute soak — BER target depends on FEC class; G.709 GFEC requires <10⁻¹² post-FEC; SD-FEC effectively zero for any soak duration.
3. Stress patterns — long sequences of identical bits (CRPAT, CJTPAT) test the receiver’s clock-recovery and equaliser tracking, complementing PRBS.
4. Far-end loopback removed, end-to-end test — replicates client-signal experience.

For coherent line-side, BERT on the OTU/ODU client interface (e.g. 100GE on a 400G coherent module) is more useful than BERT on the line side itself, because line-side errors are corrected by FEC before reaching the BERT.

FEC Margin from Coherent Telemetry

Modern coherent transponders expose pre-FEC BER and Q-margin as telemetry — these are the in-service equivalents of a BERT measurement, available continuously without a test set.

Metric	What it measures	Alarm threshold (typical)
Pre-FEC BER	Raw bit-error rate before FEC decoding	50–80 % of FEC threshold (1.5e-2 for SD-FEC)
Post-FEC BER	After FEC correction	Should be effectively 0 for any soak; non-zero indicates impending fail
Q-factor (margin)	dB above FEC threshold	<2 dB margin = warning; <1 dB = critical
OSNR (DSP-derived)	In-band OSNR	Per-modulation threshold

Key Insight

Pre-FEC BER and Q-margin are the leading indicators of impending failure. A circuit with 4 dB of Q-margin can lose 1 dB to fibre-plant aging without service impact, but the trend (e.g. losing 0.3 dB per quarter) is what predicts when remediation is needed. Streaming this telemetry into a closed-loop automation pipeline is the modern alternative to schedule-based maintenance.

Tandem Connection Monitoring (TCM)

A typical international OTN service crosses multiple administrative domains: customer LAN → carrier A’s metro → carrier A’s backbone → carrier B’s submarine → carrier C’s backbone → customer LAN at the far end. End-to-end OTN monitoring tells you “the OTN circuit is degraded”. TCM tells you which segment.

ITU-T G.709 defines six TCM levels (TCM1 through TCM6), each carried in dedicated overhead bytes. Each operator sets up a TCM endpoint at its domain ingress and egress, so the operator can monitor the performance of its own segment independently of the rest:

flowchart LR
    A["Customer A<br/>LAN"] --> CARR_A_IN["Carrier A<br/>ingress<br/>TCM1 start"]
    CARR_A_IN --> CARR_A_OUT["Carrier A<br/>egress<br/>TCM1 end"]
    CARR_A_OUT --> CARR_B_IN["Carrier B<br/>ingress<br/>TCM2 start"]
    CARR_B_IN -->|"submarine"| CARR_B_OUT["Carrier B<br/>egress<br/>TCM2 end"]
    CARR_B_OUT --> CARR_C_IN["Carrier C<br/>ingress<br/>TCM3 start"]
    CARR_C_IN --> CARR_C_OUT["Carrier C<br/>egress<br/>TCM3 end"]
    CARR_C_OUT --> Z["Customer Z<br/>LAN"]
    style A fill:#378ADD,stroke:#185FA5,color:#fff
    style Z fill:#378ADD,stroke:#185FA5,color:#fff
    style CARR_A_IN fill:#1D9E75,stroke:#0F6E56,color:#fff
    style CARR_A_OUT fill:#1D9E75,stroke:#0F6E56,color:#fff
    style CARR_B_IN fill:#7F77DD,stroke:#534AB7,color:#fff
    style CARR_B_OUT fill:#7F77DD,stroke:#534AB7,color:#fff
    style CARR_C_IN fill:#BA7517,stroke:#854F0B,color:#fff
    style CARR_C_OUT fill:#BA7517,stroke:#854F0B,color:#fff

When the end-to-end circuit degrades and customer-A escalates to carrier-A, the per-TCM error counters immediately localise the fault to one segment. Before TCM, every operator denied responsibility and the customer had to drive the dispute. With TCM, the data is unambiguous: TCM2 BIP-8 errors mean the problem is in carrier B’s segment.

Fault-Isolation Decision Tree

flowchart TD
    SYMPTOM["Service degraded or down"]
    SYMPTOM --> CLIENT{"Client<br/>signal<br/>lost?"}
    CLIENT -->|"yes"| CLIENT_PATH{"OTU layer<br/>framed?"}
    CLIENT -->|"no, but errored"| FEC_BER["Read pre-FEC BER<br/>+ Q-margin from<br/>coherent transponder"]
    CLIENT_PATH -->|"no"| LOL["Loss of light<br/>or LOF — physical layer"]
    CLIENT_PATH -->|"yes"| TCM_CHECK["Read TCM<br/>per-segment errors"]
    LOL --> OTDR_TEST["OTDR from<br/>each end<br/>locate event"]
    OTDR_TEST --> EVENT_TYPE{"Event<br/>character?"}
    EVENT_TYPE -->|"reflective spike"| CONN["Open or dirty<br/>connector"]
    EVENT_TYPE -->|"step loss, no reflection"| SPLICE["Degraded splice<br/>or fibre damage"]
    EVENT_TYPE -->|"end of trace"| BREAK["Fibre break<br/>at distance"]
    FEC_BER --> Q_MARGIN{"Q-margin?"}
    Q_MARGIN -->|"low, OSNR low"| AMP_NF["EDFA NF rise<br/>or span-loss creep<br/>OSA + EDFA telemetry"]
    Q_MARGIN -->|"low, OSNR ok"| MOD["Modulation<br/>impairment<br/>CD/PMD drift, polarisation"]
    TCM_CHECK -->|"errors in TCMx"| ISOLATE["Fault localised<br/>to that operator's<br/>segment"]
    style SYMPTOM fill:#E24B4A,stroke:#A32D2D,color:#fff
    style LOL fill:#D85A30,stroke:#993C1D,color:#fff
    style OTDR_TEST fill:#378ADD,stroke:#185FA5,color:#fff
    style FEC_BER fill:#7F77DD,stroke:#534AB7,color:#fff
    style TCM_CHECK fill:#BA7517,stroke:#854F0B,color:#fff
    style CONN fill:#1D9E75,stroke:#0F6E56,color:#fff
    style SPLICE fill:#1D9E75,stroke:#0F6E56,color:#fff
    style BREAK fill:#1D9E75,stroke:#0F6E56,color:#fff
    style AMP_NF fill:#1D9E75,stroke:#0F6E56,color:#fff
    style MOD fill:#1D9E75,stroke:#0F6E56,color:#fff
    style ISOLATE fill:#1D9E75,stroke:#0F6E56,color:#fff

T&M Tool ↔ Failure-Mode Matrix

Failure mode	OTDR	OSA	BERT	Coherent-DSP telemetry	TCM	OPM at ROADM
Fibre cut	primary — locates distance	indirect (loss of channel power)	sees errors / loss	sees LoL alarm	localises to segment	sees power loss
Splice degradation (creeping loss)	primary — bidirectional shows asymmetry	total channel power drop	post-FEC errors only at threshold	OSNR drift over weeks	localises to segment	per-channel power drift
Span-loss creep (cumulative connector / splice aging)	primary	sees OSNR drop	margin shrink	excellent — in-service trend	localises to segment	very good
Amplifier NF rise (pump aging)	no	OSNR step	margin shrink	primary — DSP-OSNR	indirect	indirect
Modulation impairment (polarisation, PMD drift)	no	no	line-side BER rise	primary — equaliser-tap drift	localises to segment	no
OTU/ODU framing fault	no	no	primary	sees LOF alarm	primary — per-segment BIP errors	no
Connector dirty / damaged	primary — reflectance	total power drop	sees errors	sees OSNR / Q-margin drop	localises	per-channel power
Spurious wavelength / drift	no	primary — sees out-of-grid signal	no	no	no	partial
Power tilt across band	no	primary — flatness measurement	no	per-channel margin variation	no	primary

OTU/ODU Service Turn-Up Checklist

A disciplined turn-up tests each layer in order — Layer 0 (optical power), Layer 1 (OTU/ODU framing), Layer 2 (client signal). Skipping a layer means a defect at a lower layer can masquerade as a higher-layer problem.

Step	Layer	Action	Pass criterion	Tool
1	L0 fibre plant	OTDR from each end of every span	All events within engineered budget; bidirectional asymmetry <0.5 dB	OTDR + launch cable
2	L0 span loss	End-to-end span loss measurement	Within ±0.5 dB of design budget	Light source + power meter
3	L0 amplifier turn-up	EDFA gain locked to span loss; AGC enabled	Per-amp gain matches measured span loss within 0.3 dB	NMS / EMS
4	L0 channel power	OPM at every ROADM degree, per-channel power	All channels within ±1 dB of target	Built-in OPM
5	L0 OSNR	Coherent-DSP-derived OSNR per circuit	OSNR ≥ required + 2 dB end-of-life margin	Transponder telemetry
6	L1 OTU framing	OTUk synchronisation lock	OTUk-LOF clear; OOF clear; A1A2 alignment	Transponder alarms
7	L1 FEC margin	Pre-FEC BER + Q-margin	Q-margin ≥3 dB; BER orders of magnitude below FEC threshold	Transponder telemetry
8	L1 OTN OH integrity	TTI (Trail-Trace Identifier), section/path BIP-8	TIM clear; BIP-8 errors zero over soak	Transponder OH read
9	L1 TCM	TCM endpoints provisioned and active	Per-TCM error counters zero over soak	NMS
10	L1 stress	PRBS31 over OTU client, ≥30 min soak	Zero post-FEC errors	BERT
11	L2 client	Client signal mapped (10GE / 100GE / etc.); FCS / framing on client	Client interface UP/UP; far-end mirror traffic	Router / switch counters
12	L2 stress	RFC2544 throughput / latency / loss test for Ethernet, ≥10 min	Zero loss at line rate; latency within budget	Traffic generator
13	Inter-layer	Verify hold-off-timer config; trigger test failures (wavelength loopback)	Layers react in correct order; no thrashing	NMS + manual induce
14	Documentation	Final OTDR traces, OSNR readout, BER soak result, configuration export	All artefacts archived per asset record	NMS export

Rule of Thumb

Always do at least one destructive failure test at commissioning — fibre-pull, transceiver-pull, ROADM-port-disable — to verify that protection / restoration actually works as designed. The number of “1+1 protected” services that have never been tested at install and quietly fail to switch when the real cut comes is depressingly large. See 14-optical-slas-availability-and-ops-reality.

See Also

• 01-optical-physics-and-link-engineering
• 03-optical-amplifiers-edfa-raman-cl-band
• 04-otn-sdh-and-network-design
• 11-optical-layer-protection-and-restoration
• 14-optical-slas-availability-and-ops-reality

References

Standards (ITU-T)

1. ITU-T G.709/Y.1331 — Interfaces for the Optical Transport Network (06/2020). https://www.itu.int/rec/T-REC-G.709
2. ITU-T G.798 — Characteristics of optical transport network hierarchy equipment functional blocks (12/2017). https://www.itu.int/rec/T-REC-G.798
3. ITU-T G.697 — Optical monitoring for dense wavelength division multiplexing systems (10/2019). https://www.itu.int/rec/T-REC-G.697
4. ITU-T G.650.1 / G.650.2 / G.650.3 — Definitions and test methods for linear, deterministic and statistical attributes of single-mode fibre and cable. https://www.itu.int/rec/T-REC-G.650.1
5. ITU-T G.661 / G.662 — Optical amplifier devices and subsystems definitions and test methods. https://www.itu.int/rec/T-REC-G.661

Standards (IEC)

6. IEC 61280-2-9 — Fibre optic communication subsystem test procedures — Optical signal-to-noise ratio measurement for DWDM systems. https://webstore.iec.ch/
7. IEC 61280-4-1 — Multimode attenuation measurement (informative for SMF practice). https://webstore.iec.ch/
8. IEC 61746 — Calibration of OTDRs. https://webstore.iec.ch/

Standards (IETF / IEEE)

9. RFC 2544 — Benchmarking Methodology for Network Interconnect Devices (03/1999). https://www.rfc-editor.org/rfc/rfc2544
10. IEEE 802.3 — Ethernet. https://standards.ieee.org/standard/802_3-2022.html

Books / industry guides

11. D. Anderson, L. Johnson, F. G. Bell, Troubleshooting Optical Fiber Networks, 2nd ed., Academic Press, 2004.
12. VIAVI / EXFO / Anritsu OTDR field-test application notes — vendor literature widely cited as practical references.
13. R. Ramaswami, K. N. Sivarajan, G. H. Sasaki, Optical Networks: A Practical Perspective, 3rd ed., Morgan Kaufmann, 2009.