How the Moon Impacts Subsea Communication Cables: Latency Variations on Transpacific Cable

Table of Links

Abstract and Introduction

GPS Long-Term Stabilized RF Phase Meter

Simple and Accurate models for tides

Latency Variations on Transpacific Cable

Poisson effect on pressurized cables

Conclusions, Acknowledgments, and References

4. LATENCY VARIATIONS ON TRANSPACIFIC CABLE

Our continuous latency recording on a modern subsea cable connecting Japan and USA started at 2020-02-28T06:06:29 UTC, ran over ~12 days, and covered a neap tide and parts of the preceding and succeeding spring tides. The geographical path of the cable is well-approximated by an arc with minimum length on the globe’s surface, which crosses areas with tidal M2 constituents of primarily low height (Fig. 2).

The probe generated in Japan was optically looped back at the other cable end in the US and the measured phase difference (MPD) to the btb signal is shown in Fig. 3a over time and relative to the aggregated tide (AT) computed with the GOT4.7.

As AT we define the integral of the local sea level elevation above normal along the cable normalized to its length. Note, at certain instants, significant local water elevations in different cable segments can cancel each other out and therefore the AT equals zero. We assume a proportionality between the local cable stretching and the corresponding water surface elevation. Hence, the overall length change linearly relates to the AT. Noise in the raw data, relating to ASE, acoustic effects, pol-scrambling, etc. were averaged over increments of 600s which suppresses high frequency components but still allows tracking of the tidal impact that appears at much lower speeds. The expected errors due to averaging are about 5%.

The MPD decomposes into a relatively strong linear tilt and a fine structure, which highly correlates with the AT. As an example, during neap tide the semidiurnal structure of both the AT and the MPD washes out, but returns at the surrounding spring tides. A linear tilt of the MPD across most of the recording does not correlate with the AT’s fine structure, but rather suggests a constant shrinking of the cable. This tilt seems to flatten towards the recording’s start and ending phases which coincides with new moon and full moon.

A variety of phenomena, like seasonal temperature drift along the cable, are currently being examined to explain its very small shrinking at a slew rate ~8 10-14/s. The ECL’s small -drift (Fig. 3c) logged every 60s (ECL~26 MHz) and averaged over 600s (mECL ~8 MHz) can be clearly ruled out as the source of the medium and long-term MPD evolution, considering the fiber CD ~21 ps/km/nm. Also, short and long-term frequency instabilities of the GPS-coupled synthesizer during the signal’s round-trip time and during the whole recording, respectively, contribute a negligible phase error. We estimate a variance for the phase error gene-rated on the Tx side in a range of Tx ~ 0.01°. No correlation has been observed between the MPD and temperature variations produced by the station’s A/C units and captured with two sensors next to our setup (Fig. 3b).