1. Introduction

An open submarine cable is bought and sold on a quantity that no single instrument reads directly: the generalized signal-to-noise ratio the wet plant delivers, stated as a function of optical frequency across the usable band. ITU-T G.977.1 names GSNR as the measurement parameter for the interoperable cable portion, and the same Recommendation requires that measurements be conducted across the usable bandwidth, because amplifier noise figure, gain shape, scattering and nonlinear interference all vary with frequency (standard-specified, ITU-T G.977.1). Channel sweep is the procedure that produces that function, one center frequency at a time.

The procedure exists because the alternative does not scale. Populating a 4.5 THz band with test transponders on every slot would need sixty coherent modems per fiber pair per direction, which is neither affordable nor available at a cable landing station during commissioning. G.977.1 therefore recommends a minimum of three test transponders plus ASE as power holders for the remainder of the optical spectrum (standard-specified, ITU-T G.977.1). The three-channel block moves; the ASE stays. Each stop yields one Q² reading that maps back to a signal-to-noise value, and the assembled stops form the profile.

Four groups act on that profile directly. Cable purchasers use it to test the delivered plant against the interoperable cable budget agreed at contract, since acceptance turns on measured average and worst-case GSNR rather than on a vendor model. Terminal equipment suppliers use it to predict capacity for a specific modem generation, because GSNR characterizes the line alone and is therefore transferable between transponder vendors. Spectrum sellers use it to price sub-bands, since a fiber pair partitioned by frequency is worth different amounts at different frequencies. Operations teams use the commissioning sweep as the baseline against which aging, repairs and gain-tilt changes are later measured, which is why the measurement conditions matter as much as the values.

The sections that follow build the procedure from its primitive. Section 2 defines what one sweep point contains and separates the four signal-to-noise quantities that are routinely conflated. Section 3 gives the noise partition and the inverse back-to-back mapping that turns a bit error ratio into a line-only metric. Sections 4 and 5 cover the physical test path and the bandwidth arithmetic. Sections 6 and 7 cover sweep planning, launch profile selection and execution. Sections 8 through 12 cover lifetime trending, fault signatures, method alternatives, roadmap and a consolidated reference set. Throughout, the boundary conditions are stated with the values: this treatment applies to dispersion-uncompensated repeatered systems where the Gaussian noise model holds, and Section 9 names where it does not.

2. Channel Sweep Definition and Component Terms

A channel sweep is a commissioning measurement in which a small block of modulated test channels is stepped in frequency across a submarine cable's usable band, with amplified spontaneous emission holding every unused slot at operational power, so that signal-to-noise performance is sampled at each center frequency in turn. The output is a table of values against frequency, not a single number.

2.1 Distinctions From Adjacent Quantities

Five quantities carry signal-to-noise names in an open-cable report, and confusing any two of them changes the answer by several decibels. Each is separated below by exactly what it includes.

• OSNR_ASE against SNR_ASE — the same linear amplifier noise, expressed in different bandwidths. OSNR_ASE is referenced to 0.1 nm (12.5 GHz at 1550 nm); SNR_ASE is referenced to the channel spacing, which removes the dependence on how many channels happen to be lit.
• SNR_ASE against GSNR — SNR_ASE counts only amplified spontaneous emission. GSNR adds nonlinear interference and guided acoustic wave Brillouin scattering, so GSNR is always the lower of the two on a lit cable.
• GSNR against SNR_EXT — SNR_EXT is what the inverse mapping returns directly, and it still contains SNR_i, the residual interaction between modem and line: polarization-dependent loss, wavelength tolerance, equalization-enhanced phase noise and imperfect dispersion compensation. Removing SNR_i in reciprocals leaves GSNR.
• GSNR against SNR_TOT — SNR_TOT adds the modem's own back-to-back noise, SNR_MODEM. GSNR describes the line and transfers between transponder vendors; SNR_TOT describes one link with one modem in it and does not transfer.
• Channel sweep against power hunt — a sweep steps the test block in frequency at a fixed launch profile and returns performance against frequency. A power hunt holds frequency and steps launch power, returning performance against power. The two answer different questions and are run in sequence, not interchangeably.