Link Budgets and Data Rates

The rate a Mars surface vehicle can achieve is set by which antenna is used, whether the far end is an orbiter overhead or a Deep Space Network dish, the Earth to Mars range at the time, and the coding. On MER the ordering of those choices spans four and a half orders of magnitude, from 10 bit/s on the low gain antenna to Earth up to 256 kbit/s on UHF to Odyssey [1].

Rates by band and path

Path	Vehicle	Rate
UHF return to orbiter	Spirit, Opportunity	8, 32, 128 or 256 kbit/s, radio capable of all four
UHF forward from orbiter	Spirit, Opportunity	8 kbit/s only, limited by the flight system design
UHF return to orbiter	Curiosity	2 to 2048 kbit/s available; Odyssey supports only 8, 32, 128, 256
UHF forward from orbiter	Curiosity	2 to 256 kbit/s; 32 kbit/s nominal with MRO, 8 or 32 with Odyssey
X-band direct to Earth, high gain antenna	Spirit, Opportunity	up to 28.8 kbit/s at short range
X-band direct from Earth, high gain antenna	Spirit, Opportunity	up to 2 kbit/s at short range
X-band direct to Earth, low gain antenna	Spirit, Opportunity	10 bit/s minimum
X-band direct from Earth, low gain antenna	Spirit, Opportunity	7.8125 bit/s minimum
X-band direct to Earth, high gain antenna	Curiosity	requirement 160 bit/s to a 34 m station, 800 bit/s to a 70 m station
X-band direct from Earth, high gain antenna	Curiosity	typically 1 or 2 kbit/s, never below 500 bit/s to a 34 m station
X-band direct from Earth, low gain antenna	Curiosity	15 bit/s at maximum range; 7.8125 bit/s at 70 degrees off boresight to a 70 m station

Sources: [1] for MER, [2] for Curiosity. The gap between the UHF and X-band columns is the argument for relay: Curiosity’s direct-to-Earth downlink requirement is 160 bit/s while its relay return link reaches about 1.35 Mbit/s.

What sets the effective rate

Antenna and pointing. Curiosity’s high gain antenna is 0.28 m across on a two degree of freedom gimbal with 5 degree system pointing accuracy including rover attitude knowledge, and at 5 degrees off boresight the downlink gain is about 4 dB lower and the uplink gain about 3 dB lower [2]. The solid state power amplifier transmits 15 W. On MER the antenna choice ordering by capability is, from least to greatest, low gain antenna to 34 m stations, low gain antenna to 70 m stations, high gain antenna to 34 m stations, and Earth-pointed high gain antenna to 70 m stations [1].

Range. MER’s nominal cruise uplink rate was 125 bit/s. On the surface the high gain antenna uplink started at 1000 bit/s and fell to 500 bit/s as the Earth to Mars distance grew; the low gain antenna uplink started at 31.25 bit/s and was later reduced to 15.625 bit/s. Curiosity shows the same dependence: its low gain antenna uplink capability dips to 15.625 bit/s at large ranges at a 40 degree off-boresight angle and to 7.8125 bit/s at 70 degrees, and is always at least 31.25 bit/s if a 70 m station is scheduled [2]. Odyssey’s own X-band downlink to the Deep Space Network started the MER primary mission at up to 110 kbit/s and fell to about 40 kbit/s into a 70 m station and 14 kbit/s into a 34 m antenna as range increased [1].

Coding. MER telemetry is convolutionally encoded at either (7,1/2) or (15,1/6), with Reed-Solomon encoding and frame and packet formatting done in avionics before the telecom hardware applies the convolutional code; the data clock is doubled for (7,1/2) and multiplied by six for (15,1/6) [1]. For a given antenna and station combination, (15,1/6) gives on average slightly greater capability than (7,1/2). The UHF relay links use bypass or (7,1/2) convolutional coding on both forward and return [2]. Perseverance’s UHF return link added low-density parity-check coding, the first on a Mars lander, and raised the bitstream mode maximum from 256 kbit/s on Curiosity to 2 Mbit/s [8]. Set against convolutionally coded overflights, LDPC returned 26 percent more data on the Trace Gas Orbiter link and about 70 percent more, equivalently 2.3 dB, on the MAVEN link over the first 27 LDPC overflights [7]. The MAVEN figure is estimated against convolutional predicts rather than paired passes, and the gain varies with orbit geometry.

Modulation and adaptive rate. Proximity-1 carries the data rate as a negotiable session parameter rather than a fixed configuration, which is what makes an in-pass rate change possible at all [4]. The Curiosity to MRO return link uses suppressed carrier modulation with adaptive data rates, changing the return rate during a pass under Proximity-1 control; Odyssey links must use residual carrier at a fixed rate for the whole contact [2]. Only the return rate is adapted, the forward rate staying at 32 kbit/s, and the hailing interaction that opens every link runs at 8 kbit/s. Throughput efficiencies at the MRO or MSL end cap the effective maximum around 1.35 Mbit/s, depending on how much forward data MRO is sending at the same time.

Measured against predicted

Odyssey’s best recorded MER pass returned 170 Mb, while the best UHF pass at 128 kbit/s was on the order of 110 Mb; a horizon to horizon Odyssey overflight can last as long as 17 min, and the 120 Mb per rover buffer allocation corresponds to about 15 min of data at 128 kbit/s [1]. Late in the extended mission at near maximum Earth range, two consecutive rover passes to Odyssey could fall only 2 hours apart while Odyssey’s minimum X-band downlink of 14 kbit/s to the Deep Space Network let it clear only about 50 Mb per hour, including higher priority Odyssey data, so relay data could be overwritten before it was sent.

Prediction error for Phoenix through MRO over a 30 day period, at a 2 dB margin policy [3]:

Elevation mask	Average data volume delta	Passes more than 10 percent below predict
10 degrees	-0.22 percent	25.00 percent
15 degrees	+13.13 percent	3.75 percent

Prediction accuracy is limited by the rover UHF antenna pattern rather than by the link equation. MER operations continued to have difficulty predicting return volume because the antenna gain pattern is asymmetric in azimuth and can be badly distorted by the vehicle structure, and the prediction packages in use were described as crude [1]. The same problem motivated building data volume prediction tools for the Phoenix and MSL surface missions [3].

One link measured end to end: MSL entry, descent and landing

The Mars Science Laboratory entry link is a rare case where a prediction, a threshold and a flight measurement of the same link are all published together. Because Doppler dynamics through banking and parachute deployment make phase-modulated telemetry impossible, MSL radiated a semaphore instead: 256 multiple-frequency-shift-keyed tones, 10 s each, on a residual carrier, the first tone at a 2000 Hz offset with 70.5883 Hz nominal spacing [6]. Canberra received it on the 70 m DSS-43 as prime and on a DSS-34 and DSS-45 34 m array as backup.

Measured. Averaged over the whole EDL pass on 5 August 2012, carrier power to noise spectral density on DSS-43 was 27.8 dB-Hz and tone power to noise density 26.4 dB-Hz [6]. Predicted before flight, from the detection-probability equations with the parachute segment parameters: about 23 dB-Hz total power to noise density for 99.9 percent carrier acquisition and about 28 dB-Hz for tone detection, both read off plotted curves rather than tabulated. All radiated tones were correctly detected in real time, verified afterwards against the spacecraft’s own record of which tones it issued [6]. That 100 percent is the design target and it arrives with no margin measurement: the paper reports no missed or marginal tone, so the operating margin above threshold is inferred from a pass average rather than demonstrated. By the authors’ own configuration reasoning some tones were expected to be missed during parachute deployment, because the receiving assembly’s 1 s minimum incoherent integration exceeds the 0.73 s the dynamics called for [6], and that prediction is never revisited against the flight data.

Arraying against a single large aperture. The 34 m array was predicted to sit 4.44 dB below the 70 m antenna, computed from G/T assuming no combining loss [6]. The measured mean difference over the window from 645 s before entry to 299 s after was 4.26 dB, implying about 0.18 dB of mean combining loss. Read that as a residual rather than a clean measurement: the prediction takes a 34 m G/T of 18 percent of a 70 m from the DSN Telecommunications Link Design Handbook rather than measuring it, so the 0.18 dB inherits the handbook value [6].

Where the link went away. Two configuration changes produced outages, 7 s at cruise stage separation where the stage briefly obstructed the Parachute Low Gain Antenna, and 4 s at the swap from that antenna to the Tilted Low Gain Antenna [6]. The closed-loop receivers additionally lost lock at cruise stage separation, the antenna swap, banking and parachute deployment, which is the reason the open-loop semaphore chain was prime rather than backup.

Event timing against the six degree of freedom prediction: bank 1 about 2 s late, bank 2 about 6 s late, bank 3 about 2 s early, maximum deceleration about 9 s late, and entry itself 1 s later than the navigation predict [6]. The source table’s predicted and actual columns transpose on at least two rows, so the individual times are unreliable.

One number on this link is a fit and not a validation. The Small Deep Space Transponder auxiliary oscillator frequency against temperature was fitted as a cubic with R squared 0.9977 against the oscillator sensor and 0.9984 against the baseplate sensor, but the fit was made to cruise data in thermal equilibrium, and EDL is rapid heating with the baseplate overshooting the oscillator [6]. No residual against the EDL data is reported, so the polynomial describes cruise and does not predict the entry excursion.

Direct to Earth against relay

Direct to Earth survives because it does not depend on another spacecraft. Curiosity’s X-band direct-from-Earth uplink is the primary path for the daily command load, about 225 kbit in 15 to 20 minutes on the high gain antenna, equivalent to an average 190 bit/s, with a 15 minute setup allowance before the window and a 10 minute rover activity keep-out afterward for post-pass processing [2]. Direct-to-Earth downlink is used for limited amounts of data independent of the relay orbiters, and the two are often combined into one station pass long enough to verify the uplinked commands in the telemetry that follows, accounting for round trip light time. On MER, in the extended mission, a flight software update was uplinked at 1000 bit/s and then at 2000 bit/s in February 2005, the highest uplink rate the 20 kW transmitters supported [1].

Contingency inverts the ordering. The MER fault configuration drops the uplink to 7.8125 bit/s and the downlink to 10 bit/s on the low gain antenna, which is the state Spirit was found in during the sol 18 flash anomaly [1].

For comparison, a link with bandwidth to spare

Delay tolerant networking on the International Space Station runs over a Ku-band space-to-ground link nominally 20 Mbit/s down and 4 Mbit/s up, with a maximum allowable 90 Mbit/s down for DTN traffic although it is typically limited to about 30 Mbit/s [5]. Measured DTN throughput was about 15 Mbit/s down and 4 Mbit/s up with 100,000 byte bundles. Those rates are three orders of magnitude above any Mars surface link, so protocol overhead and congestion control results from low Earth orbit do not carry across.