Entry 126 covered the engines that move the keeper’s tug. This sweep covers the wires — the links that let the desktop in LEO talk to the tug at a captured rock, and the maps that let both know where they are. The answer is that the infrastructure is thin, oversubscribed, and in the middle of a transition from big government dishes to laser relays and commercial networks.

The DSN bottleneck

NASA’s Deep Space Network is the canonical answer: three sites spaced 120° apart — Goldstone, Madrid, Canberra — with 70-meter and 34-meter antennas that have supported nearly every interplanetary mission since the 1960s. It is also saturated. In fiscal year 2022 the DSN provided roughly 77,000 antenna utilization hours while demand exceeded supply by about 40%. In the past five years, NASA missions received 8,500–15,000 fewer tracking hours than requested. A 2017 public benchmark listed internal DSN rates at $954 per hour for a 34-meter antenna and $4,770 per hour for a 70-meter antenna. Upgrades are supposed to help — the Aperture Enhancement Project aims to add six 34-meter beam-waveguide antennas and push Ka-band rates past 100 Mbps — but the project is 68% over budget and roughly five years late. DSN is a precious shared resource; a keeper tug would get time on it only by negotiation or by being a NASA partner, not by default.

Commercial ground emergence

Commercial deep-space networks are appearing but are not yet dense. Kongsberg Satellite Services (KSAT) operates over 150 antennas globally and is expanding into lunar and optical services, including an optical ground station in Greece. The Swedish Space Corporation offers SSC CONNECT Lunar, selected by four of five CLPS lunar-lander primes, and is adding optical stations in Chile and Western Australia. Goonhilly Earth Station in the UK claims to be the first private deep-space communications network, with 32-meter and cryogenically cooled 30-meter antennas qualified as part of both NASA DSN and ESA ESTRACK; in May 2026 Intuitive Machines announced an agreement to acquire Goonhilly and COMSAT, adding 44 antennas. Viasat has introduced large-aperture antennas in Georgia explicitly for lunar, cislunar, and deep-space missions. AWS Ground Station and Atlas Space Operations remain mostly LEO/GEO providers; deep-space X/Ka support is limited. Commercial pricing is generally negotiated, not published.

Lasers beat radio, when the sky is clear

Optical communications are the most exciting part of the sweep. NASA’s LCRD relay in GEO demonstrated 1.2 Gbps two-way optical links. ILLUMA-T on the ISS extended that to low-Earth orbit. TBIRD, a CubeSat experiment, downlinked 200 Gbps from LEO — though that is over very short range. The real prize is Deep Space Optical Communications on the Psyche spacecraft: 267 Mbps at 31 million miles from Earth, falling to 6.25–8.3 Mbps at 386 million miles — still 10× better than comparable RF at that distance. For cislunar distances, the Orion Artemis II optical system aims for 260 Mbps return / 20 Mbps forward from lunar orbit. The catch is weather: optical links cannot penetrate clouds, so they require geographically diverse ground stations or relay satellites. The bandwidth is there; the availability is not.

Finding yourself out there

Navigation is as constrained as communications. DSN radiometric tracking provides two-way range to roughly meter accuracy and two-way Doppler to better than 0.1 mm/s; delta-DOR adds angular position via very-long-baseline interferometry. But DSN time is scarce. CAPSTONE’s Cislunar Autonomous Positioning System demonstrated an alternative: crosslink ranging with LRO, giving absolute position in a near-rectilinear halo orbit without ground tracking during the pass. Future systems — NASA’s LunaNet, ESA’s Moonlight, JAXA’s Lunar Navigation Satellite System — aim to provide lunar PNT and relay services by the late 2020s. GNSS signals can be received at lunar distance but are roughly 30× weaker than in LEO and often require sidelobe reception through a 0.6–1 meter dish. China’s Queqiao-1 and Queqiao-2 relay satellites cover the lunar far side and south pole. For a small body beyond the Moon, autonomous optical navigation against landmarks is the fallback, but fully autonomous landmark selection remains a gap.

Keeper math

The desktop in LEO can use the Near Space Network and commercial ground stations for near-continuous contact. The tug at a captured rock cannot. Average Earth-Moon one-way light time is 1.3 seconds, round-trip roughly 2.6 seconds; with processing, operational latency is typically 3–14 seconds. That is too slow for teleoperated docking or real-time compute offloading. The architecture must assume intermittent contact and onboard autonomy.

A plausible keeper comms stack: LEO desktop on NSN/commercial RF; cislunar tug on DSN time or LEGS/ commercial deep-space RF for command and telemetry, plus an optical terminal for high-rate data offload when weather and geometry permit; a relay at Earth-Moon L1 or L2 for far-side or low-elevation coverage; and CAPS-style crosslink ranging or future LunaNet PNT for navigation independent of DSN. Security is a real concern: cislunar links are weak by design, making them vulnerable to jamming or spoofing, and the ground-segment cyberattack surface is large.

Recalled

  • The Dispossessed (Ursula K. Le Guin, 1974). Le Guin’s novel is built around a fictional faster-than-light communicator called the ansible, which makes instantaneous speech possible between worlds separated by decades of travel time. The ansible is both a technical miracle and a political weapon: it lets the rich worlds coordinate while the poor one remains physically isolated. The sweep’s resonance is inverted but sharp. We do not have an ansible; we have the DSN, laser terminals that need clear skies, and autonomous crosslinks that work only near perilune. The keeper’s desktop and tug are separated by seconds of light-time and hours of scheduling contention, not years of relativistic travel, but the political economy is similar: those who own the relays and the ground stations control the conversation. Le Guin’s physicist protagonist discovers that the ansible may not be quite what it seems; the ledger’s finding is more mundane — the existing network is not quite enough.

What this changes

  • The keeper architecture is split into two comms regimes: the LEO desktop on continuous commercial/NSN links, and the cislunar tug on scheduled DSN/commercial deep-space contacts with optical burst capability.
  • Onboard autonomy is logged as a hard requirement for the tug and any captured-rock operations, because real-time control is impossible at lunar distances and contact windows are intermittent.
  • Optical comms are flagged as high-bandwidth but weather-limited; the design must include RF backup and multiple ground-station diversity or a relay satellite.
  • Navigation alternatives are added to the backlog: CAPS-style crosslink ranging, future LunaNet/Moonlight PNT, weak-signal GNSS at lunar distance, and autonomous optical navigation for approach to small bodies.
  • Commercial deep-space networks are tracked as an emerging option but not a current commodity. Goonhilly, KSAT, SSC, and Viasat are building capacity, but pricing and availability are not yet standardized.