Communications

A surface vehicle has to return data and receive commands from a radio that is small, low in power and often out of view of both Earth and any relay. The software problem is to schedule, frame, store and retransmit data across links that exist for minutes per sol, at rates that differ by orders of magnitude between the antennas available, and with a round-trip light time to Mars of 10 to 40 minutes [15]. Three pages below cover the subject: link budgets and data rates, protocols and standards, and relay architecture.

Why the designs converged on relay

A direct-to-Earth link from a small antenna is limited by range and transmit power, so the flown Mars surface missions moved the bulk of their traffic to a UHF hop to an orbiter. About 95 percent of MER science data was returned through orbiters [15]. Phoenix flew without an X-band direct-to-Earth system, a change the project estimated saved 14.8 kg [2], and returned 38,120 Mb in 157 sols over 860 UHF passes, an average of 242 Mb/sol against a requirement of 60 Mb/sol [2]. The orbit choice that makes this work was argued before any rover flew: a sun-synchronous frozen relay orbit gives repeatable lander visibility, and a 10 Mb per lander per sol return was the sizing requirement [8]. The Chinese lunar program reached the same architecture with the Queqiao relay satellite, which sets the Yutu-2 commanding cadence through its forward and return rates [6].

Link budgets are built the way deep space telecommunications has done since the design control table: each parameter carries a design value with favorable and adverse tolerances, and margin is computed against clear and weather-degraded conditions [9]. The Mars missions’ pre-landing predictions and in-flight results are compared in the MSL and MER telecommunications records [24][1].

Relay allocation is an operations problem as much as a link problem. The Mars Sample Return campaign concept assumes three co-located surface assets at Jezero sharing one principal orbiter relay, and a JPL allocation tool was built to divide passes among them [26]. The Curiosity team’s 2019 solar conjunction planning negotiated buffers, data rates and blackouts with four orbiters and the DSN, and flew two templated 7-sol plans uplinked in advance [27]. The same relay logic reaches down to a 14 g helicopter telecom subsystem, whose 914 MHz Zigbee link to Perseverance uses antennas under 4 g [25].

Critical events and flight validation

Entry, descent and landing is the case where the link cannot be rescheduled. For MSL, MRO acquired the UHF link 8 min 7 s before entry interface, and plasma attenuation of the UHF signal lasted about 50 s [3]. The direct-to-Earth tone link averaged 27.8 dB-Hz carrier power to noise density at the 70 m station and every radiated tone was detected in real time [4]. For ESA’s Schiaparelli, NASA orbiters carried an open-loop recording of the 8 kb/s entry telemetry, and a radio telescope array was used to detect the carrier, which was lost 72 s before the predicted touchdown [5]. Perseverance relay planning measured the coding gain of LDPC on an operational relay link, which is the only such figure in the corpus [7].

Protocols and delay tolerance

The stack that flew on Mars orbiters is Proximity-1 for the relay hop, with CCSDS packet and AOS data link layers above and beside it [11][21]. Proximity-1 carries the session rules that EDL work uses in unreliable bitstream and open-loop modes [11]. Reliable transfer above the link layer is provided by the CCSDS File Delivery Protocol, with unacknowledged and acknowledged classes [22].

Delay and disruption tolerant networking is the store-and-forward alternative. The Bundle Protocol standard requires nodes to have a reliable storage service [12]. Simulation over real Mars relay pass schedules gave an average lander-to-operations bundle latency of 2.4 h with a 20 h maximum under first-contact routing and unlimited buffers [15]. The flown evidence is two experiments: the Deep Impact link delivered about 300 images over 27 days with a measured path utilization of 76.2 percent against a criterion above 90 percent [13], and the ISS gateway reached 15 Mbps effective downlink with no data lost during a simulated 10 minute loss of signal [14]. Integrating a DTN stack into flight software framework F Prime exposed a blocking API and pre-allocated memory as the two integration costs [23]. No planetary surface robot has flown DTN as its primary path.

Surface and cislunar links

Surface-to-surface radio has been studied more than flown. Ray-tracing over Jezero terrain gave path loss exponents from 1.93 to 2.84 and shadowing deviations from 5.3 to 16.3 dB between two sites [18]. An optical lunar surface terminal was analyzed for 200 Mb/s to over 1 Gb/s from a 1 W terminal, with the observation that simulant abrades optical surfaces [16]. NASA’s plan to baseline 3GPP cellular plus Wi-Fi for lunar surface communications included a 4G LTE system flown on IM-2 to demonstrate lander-to-rover links at 200 to 300 m and up to 2 km [28]. The draft LunaNet specification indicates that commercial terrestrial standards, rather than bespoke links, are the intended lunar surface interface [17]. Tabulated Mars lander and orbiter transceivers show the international relay interoperability that Proximity-1 makes possible [20].

Operations and human factors

Delay sets operator behavior: manipulators under transmission delay are driven by move-and-wait, and completion time scales with delay [10]. In Artemis LTV simulator studies, average teleoperated speed fell from 3.24 kph at 0 s to 1.76 kph at 8 s of round-trip delay [29]. The Deep Space Network moved from 14 operators per antenna in 1980 to 4 by 1990 and from one link per operator to two in 2012 [19].

What is not established

Surface mesh propagation on Mars rests on simulation, not measurement [18]. Optical surface links and lunar DTN are analysis or ground test only [16][15]. DTN has no flight record on a planetary relay path. The relay planning results come from one rover and one orbiter fleet, so the transfer to other relays is not demonstrated [7].