Chapter 26: Communications, Navigation, and Data Handling

"The fundamental problem of communication is that of reproducing at one point either exactly or approximately a message selected at another point." — Claude Shannon, A Mathematical Theory of Communication (1948)

Overview

A spacecraft that cannot talk to Earth is not a spacecraft. It is a very expensive rock on a very precise trajectory, and no one will ever know what it saw. Everything a probe is for — the images, the spectra, the magnetometer readings, the confirmation that the burn worked — has to come home as a radio signal, and every command that keeps it alive has to go out the same way. This chapter is about that thread of radio, thinner and more fragile than almost anyone imagines, and about the two things that ride on it besides data: navigation (the same radio link tells us where the spacecraft is and how fast it is going, to a precision that beggars belief) and the onboard computer that decides what to send.

Here is the physical heart of it, and it is worth feeling before we quantify it. New Horizons flew past Pluto in July 2015 with a radio transmitter of about 12 watts — less than a refrigerator's interior light bulb. By the time that signal reached Earth, spread across roughly $4.8$ billion kilometers of empty space, the antenna that caught it collected about one and a half billionths of a billionth of a watt ($1.4\times10^{-18}\ \text{W}$). And out of that whisper, engineers pulled sharp pictures of a world no one had ever seen. This is the recurring lesson of Part IV, and of this chapter especially: space is unforgiving, and it gets more unforgiving with distance. The signal weakens as the square of how far it has traveled — the same inverse-square law that governs gravity and sunlight — and there is no way to cheat it. You can only build for it: bigger antennas, colder receivers, cleverer codes, and the patience to receive Pluto at the speed of a 1990s dial-up modem.

In Chapter 25 you learned to size the power a spacecraft generates; the radio transmitter is very often the single hungriest load on that power budget, so the two chapters are joined at the hip. Here we learn where that power goes once it leaves the antenna, how little of it comes back, and why that is enough.

In this chapter, you will learn to:

• Build a link budget in decibels and read off whether a link "closes" — whether the signal beats the noise.
• Compute the gain and beamwidth of a dish antenna, and see why pointing a deep-space antenna is hard.
• Explain what the Deep Space Network is, why it has exactly three main sites, and what its 70 m dishes buy.
• Relate frequency band, modulation, and error-correcting codes to the data rate a mission can achieve.
• Track a spacecraft to the meter and the millimeter-per-second using ranging and Doppler, and to the kilometer across the sky using delta-DOR.
• Say why the computer flying a billion-dollar probe is slower than the phone in your pocket, on purpose.

Learning Paths

🚀 Space Enthusiast: Read 26.1 for the idea of a link budget (you can skim the decibel algebra), then enjoy 26.3 (the Deep Space Network) and 26.5 (how we navigate across the solar system). The "whisper in a hurricane" reality of 26.1 and the nanoradian precision of delta-DOR in 26.5 are the two facts that will change how you read a mission press kit.

📐 Engineering Student: Read everything and do the New Horizons link budget in 26.1 yourself before reading ours, working in dB throughout. The link-budget equation and the antenna-gain formula in 26.2 are the load-bearing tools; the ⭐⭐/⭐⭐⭐ exercises build the fluency you will need in Chapter 27.

🎮 KSP Player: Your relay dishes and the "connection lost" warning are a cartoon of this chapter. Focus on 26.1 (why range matters so much), 26.3 (why you need relays on the far side of a body), and 26.4 (why bigger antennas and higher bands mean more science per pass).

🛰️ Industry Prep: The link budget of 26.1 and the band/coding trades of 26.4 are exactly what a communications systems engineer defends at a design review. You will write your mission's comms note — band, data rate, ground stations — into your Mission Design Review at the end of this chapter.

26.1 The link budget

Every radio link, from your phone to a Wi-Fi router to New Horizons at Pluto, obeys one accounting identity: the power that arrives equals the power that left, times everything that helped (antenna focusing), divided by everything that hurt (the spreading of the signal through space, and various losses). That accounting is called a link budget, and mastering it is most of this chapter.

Definition (link budget). A link budget is the tally of all the gains and losses a communication signal experiences between transmitter and receiver, added up to predict the received signal power and the resulting signal-to-noise ratio. A link "closes" when the received signal-to-noise ratio exceeds the minimum the receiver needs to recover the data.

Why we work in decibels

The numbers in a deep-space link span an absurd range: a 12 W transmitter, a path loss that divides the signal by a factor of $10^{30}$, an antenna gain that multiplies it back by $20{,}000$. Multiplying and dividing quantities that big is error-prone, so radio engineers do the whole calculation in decibels, where multiplication becomes addition.

A decibel expresses a power ratio on a logarithmic scale. A power $P$ relative to a reference $P_{\text{ref}}$ is

$$ P\,[\text{dB}] = 10\,\log_{10}\!\left(\frac{P}{P_{\text{ref}}}\right). $$

When the reference is 1 watt we write dBW; when it is 1 milliwatt, dBm. A gain (a pure ratio, like an antenna's) is written in dB, and when it is measured relative to an isotropic radiator — an idealized antenna that radiates equally in all directions — we write dBi. The rules are just the laws of logarithms: to multiply two quantities, add their decibels; to divide, subtract.

💡 Intuition: Decibels turn the link budget into simple bookkeeping. Instead of $P_r = P_t \times G_t \times G_r \div L$, a product of a dozen huge and tiny factors, you write a single column of numbers and add them up: start with the transmitter power in dBW, add each gain in dB, subtract each loss in dB, and the bottom line is the received power in dBW. A link budget is literally a budget — a column of credits and debits — which is why it is drawn as a table.

A few conversions worth internalizing: every factor of 10 in power is 10 dB; every factor of 2 is very close to 3 dB (because $10\log_{10}2 = 3.01$); a factor of 1000 is 30 dB. So 12 W is $10\log_{10}(12) \approx 10.8\ \text{dBW}$, and a 200,000× antenna gain is $\approx 53\ \text{dB}$.

The free-space path loss

Now the physics. A transmitter radiates power $P_t$. If it did so isotropically, that power would spread uniformly over the surface of an expanding sphere, so at distance $d$ the power flux density (watts per square meter) would be

$$ S = \frac{P_t}{4\pi d^2}. $$

There it is — the inverse-square law, identical in form to the one governing gravity in Chapter 2 and sunlight in Chapter 24. Double the distance and the flux drops by a factor of four. A real antenna concentrates its power in a preferred direction, multiplying the on-axis flux by its gain $G_t$, so $S = P_t G_t / (4\pi d^2)$. The receiving antenna then intercepts this flux over its effective area $A_e$, collecting $P_r = S\,A_e$. Antenna theory (next section) gives the effective area in terms of the receive gain, $A_e = G_r \lambda^2/(4\pi)$, where $\lambda$ is the wavelength. Put the pieces together:

$$ P_r = \frac{P_t\,G_t}{4\pi d^2}\cdot\frac{G_r\,\lambda^2}{4\pi} = P_t\,G_t\,G_r\left(\frac{\lambda}{4\pi d}\right)^2. $$

This is the Friis transmission equation, the master equation of every radio link. The factor $(\lambda / 4\pi d)^2$ — or rather its reciprocal — is given a name of its own.

Definition (free-space path loss). The free-space path loss is the factor by which signal power is reduced traveling a distance $d$ between two isotropic antennas at wavelength $\lambda$: $$L_{\text{fs}} = \left(\frac{4\pi d}{\lambda}\right)^2, \qquad L_{\text{fs}}\,[\text{dB}] = 20\,\log_{10}\!\left(\frac{4\pi d}{\lambda}\right).$$ It combines the inverse-square spreading of the signal with the wavelength-dependent aperture of an ideal receiving antenna. Note the $20\log_{10}$: the quantity inside is squared, and $10\log_{10}(x^2) = 20\log_{10}(x)$.

With everything named, the Friis equation in decibels becomes a sum — the link-budget equation we will use for the rest of the chapter:

$$ \boxed{\ P_r = P_t + G_t + G_r - L_{\text{fs}} - L_{\text{other}}\ } $$

where every term is in decibels ($P_r, P_t$ in dBW; $G_t, G_r$ in dBi; the losses in dB), and $L_{\text{other}}$ collects the smaller losses — atmospheric absorption, imperfect antenna pointing, polarization mismatch, and receiver implementation. The sum $P_t + G_t$ is used so often it has its own name, the effective isotropic radiated power (EIRP): the transmitter power an isotropic antenna would need to produce the same on-axis signal.

🚪 Threshold Concept: the tyranny of $d^2$ in communication. In most engineering, "twice as far" is twice as hard. In radio it is four times as hard, because the received power falls as the square of distance — and this is the single fact that shapes every deep-space communications system. It is why a probe at Pluto (32 AU) receives $(32/5)^2 \approx 40$ times less signal than the same probe did at Jupiter (5 AU), and returns data forty times slower. It is why the antennas that hear these probes are the size of baseball diamonds. It is why the mass and power a mission spends on communications climbs relentlessly the farther it goes. You cannot argue with the inverse-square law; you can only build a bigger dish, cool a quieter receiver, and slow down. Once you see the $d^2$, you understand why the outer solar system is quiet — not because the probes are weak, but because space is vast.

Worked example: the New Horizons downlink from Pluto

Let us do the calculation that opened the chapter. All figures are widely reported, approximate values (Tier 2), rounded for legibility; the point is the method and the order of magnitude.

Worked Example: does the New Horizons link close at Pluto?

The setup. New Horizons carries a $D = 2.1\ \text{m}$ high-gain dish and transmits about $P_t = 12\ \text{W}$ in the deep-space X-band, at a downlink frequency of about $f = 8.4\ \text{GHz}$. At Pluto encounter it was about $d = 4.83\times10^{12}\ \text{m}$ from Earth (≈ 32 AU, ≈ 3 billion miles). On the ground, a Deep Space Network 70 m dish receives the signal. First, the wavelength: $\lambda = c/f = (2.998\times10^8)/(8.4\times10^9) = 0.0357\ \text{m}$ (3.57 cm).

Transmitter power. $P_t = 10\log_{10}(12) = +10.8\ \text{dBW}$.

Spacecraft antenna gain. Using the dish-gain formula of Section 26.2 with efficiency $\eta = 0.6$, $G_t = 10\log_{10}\!\big[\eta(\pi D/\lambda)^2\big] = +43.1\ \text{dBi}$. So the spacecraft's EIRP $= 10.8 + 43.1 = +53.9\ \text{dBW}$.

Free-space path loss. This is the brutal term: $$L_{\text{fs}} = 20\log_{10}\!\left(\frac{4\pi d}{\lambda}\right) = 20\log_{10}\!\left(\frac{4\pi (4.83\times10^{12})}{0.0357}\right) = 20\log_{10}(1.70\times10^{15}) = +304.6\ \text{dB}.$$ Read that as a power reduction of $10^{30.46}$ — thirty and a half orders of magnitude. That one number is the whole difficulty of deep space in a nutshell.

Receive antenna gain. The 70 m dish at X-band, with $\eta = 0.7$, has $G_r = +74.2\ \text{dBi}$.

Other losses. Atmosphere, pointing, and implementation together, call it $L_{\text{other}} = 2\ \text{dB}$.

Received power. Add the column: $$P_r = \underbrace{53.9}_{\text{EIRP}} + \underbrace{74.2}_{G_r} - \underbrace{304.6}_{L_{\text{fs}}} - \underbrace{2.0}_{L_{\text{other}}} = -178.5\ \text{dBW}.$$ Converting back, $-178.5\ \text{dBW} = 10^{-17.85}\ \text{W} = 1.4\times10^{-18}\ \text{W}$ — the attowatt whisper from the overview. (If you counted the individual radio photons arriving, at $hf \approx 5.6\times10^{-24}\ \text{J}$ each, it works out to about 250,000 photons per second — a firehose of photons, each one carrying almost nothing.)

Received power alone does not tell you whether the link works. What matters is the signal relative to the noise competing with it. Every receiver adds noise, characterized by a system noise temperature $T_s$ (in kelvin): the colder the receiver, the less noise, which is why the DSN cools its amplifiers to cryogenic temperatures. The noise power in each hertz of bandwidth — the noise spectral density — is

$$ N_0 = k\,T_s, \qquad k = 1.38\times10^{-23}\ \text{J/K (Boltzmann's constant)}. $$

For a DSN X-band system, $T_s \approx 20\ \text{K}$, so $N_0 = (1.38\times10^{-23})(20) = 2.76\times10^{-22}\ \text{W/Hz}$, or in decibels $N_0 = -215.6\ \text{dBW/Hz}$. (In dB this is Boltzmann's constant, the famous $-228.6\ \text{dBW/K/Hz}$, plus $10\log_{10}T_s$.) The ratio of received power to noise density is the link's fundamental figure of merit:

$$ \frac{P_r}{N_0} = -178.5 - (-215.6) = 37.1\ \text{dB-Hz}. $$

This $P_r/N_0$ — in decibel-hertz — is a rate: it is essentially how many bits per second the link can carry, once you decide how much energy each bit needs. Digital receivers quantify that need as $E_b/N_0$, the energy-per-bit to noise-density ratio, and it relates to the data rate $R$ by the tidy identity $P_r = E_b R$ (received power is energy-per-bit times bits-per-second). In decibels:

$$ \frac{E_b}{N_0} = \frac{P_r}{N_0} - 10\log_{10} R. $$

At $R = 1\ \text{kbit/s} = 1000\ \text{bit/s}$, which costs $10\log_{10}(1000) = 30\ \text{dB-Hz}$, the link delivers $E_b/N_0 = 37.1 - 30 = 7.1\ \text{dB}$. With the error-correcting coding of Section 26.4, a receiver needs only about $2.5\ \text{dB}$ to recover essentially error-free data — so the link has a margin of about $7.1 - 2.5 = 4.6\ \text{dB}$. Positive margin means the link closes: New Horizons can send about a kilobit per second from Pluto, and it did. To go faster, it would need more $P_r/N_0$ — a bigger dish on the ground (arraying, Section 26.3), which is exactly how the fastest Pluto downlinks reached ~2 kbit/s.

🔧 Engineering Reality: Notice what set the data rate: not the transmitter, not cleverness, but geometry. New Horizons could have carried a 24 W transmitter instead of 12 W and gained only 3 dB — one measly doubling of rate — at a large cost in power it did not have, because its plutonium generator produced only a couple of hundred watts total. Doubling the dish diameter, by contrast, buys 6 dB at each end. This is why deep-space engineering pours its effort into antennas and receivers, not raw wattage: near the inverse-square wall, the levers that matter are the ones that appear squared in the Friis equation.

🔄 Check Your Understanding 1. A link budget gives $P_r/N_0 = 40\ \text{dB-Hz}$ and the receiver needs $E_b/N_0 = 2\ \text{dB}$. What is the fastest data rate (in bit/s) the link can support? 2. You move a probe from 10 AU to 30 AU. By how many decibels does the free-space path loss increase, and by what factor does the received power fall?

Answers

1. The available $E_b/N_0$ margin lets you spend $40 - 2 = 38\ \text{dB-Hz}$ on data rate, so $R = 10^{38/10} = 10^{3.8} \approx 6{,}300\ \text{bit/s}$. 2. Distance triples, so path loss rises by $20\log_{10}(3) = 9.5\ \text{dB}$, and received power falls by a factor of $3^2 = 9$. The inverse-square law again — tripling the distance is nine times harder, not three.

26.2 Antennas and gain

The link budget leaned twice on the same idea — that an antenna concentrates power in a direction — so let us make it precise. The quantity is antenna gain.

Definition (antenna gain). An antenna's gain $G$ is the factor by which it increases the power flux density in its preferred (boresight) direction, compared with an isotropic radiator emitting the same total power. It is a pure ratio, expressed in dBi (decibels relative to isotropic). Gain describes both transmitting and receiving: a high-gain antenna concentrates transmitted power into a narrow beam and, used in reverse, collects incoming power from that same narrow beam.

For the parabolic dish that dominates spaceflight — a reflector that focuses a plane wave to a point, or a point source into a plane wave — the gain follows from the dish's area and the wavelength:

$$ G = \eta\left(\frac{\pi D}{\lambda}\right)^2, $$

where $D$ is the dish diameter, $\lambda$ the wavelength, and $\eta$ the aperture efficiency (how uniformly and cleanly the dish is illuminated — typically 0.5 to 0.7; it is never 1 because of spillover, blockage, and surface imperfections). The physics behind the formula is worth stating in words: gain grows as the square of the dish diameter measured in wavelengths. A dish that is many wavelengths across can form a very tight beam; a dish comparable to a wavelength can barely form a beam at all. This is why radio astronomers and deep-space engineers build enormous dishes, and why moving to a shorter wavelength (higher frequency) raises the gain of a fixed dish — a lever we will pull in Section 26.4.

Strategy first. To find a dish antenna's gain, count how many wavelengths fit across it ($\pi D/\lambda$ is the circumference in wavelengths), square that, and knock it down by the efficiency. To find its beamwidth, use the reciprocal relationship: the more wavelengths across the dish, the narrower the beam.

The flip side of high gain is a narrow beam. A useful engineering approximation for the half-power (−3 dB) beamwidth of a dish — the angular width over which the gain stays within a factor of two of its peak — is

$$ \theta_{3\text{dB}} \approx 70^\circ\,\frac{\lambda}{D} \quad (\lambda, D \text{ in the same units}), $$

and gain and beamwidth are two views of the same fact, linked by $G \approx 29{,}000/\theta_{3\text{dB}}^2$ (with $\theta$ in degrees). Squeeze the beam and the gain rises; both are set by $D/\lambda$.

Worked Example: the New Horizons dish — gain and beamwidth. With $D = 2.1\ \text{m}$ and $\lambda = 0.0357\ \text{m}$ at X-band, the dish spans $\pi D/\lambda = \pi(2.1)/0.0357 = 185$ wavelengths around its rim. Its gain (with $\eta = 0.6$) is $G = 0.6\times(185)^2 = 20{,}500$, or $10\log_{10}(20{,}500) = 43.1\ \text{dBi}$ — the number we used in the link budget. Its beamwidth is $\theta_{3\text{dB}} \approx 70^\circ\times 0.0357/2.1 = 1.2^\circ$. Cross-check with the gain–beamwidth rule: $29{,}000/(1.2)^2 = 20{,}100 \approx 43.0\ \text{dBi}$. ✓ The two formulas agree, as they must.

That $1.2^\circ$ beam is the crux of a hard engineering problem. From 4.8 billion kilometers, the spacecraft must point its beam so that Earth falls inside a cone barely a degree wide — a pointing job for the attitude-control system of Chapter 14, and one reason deep-space craft carry star trackers accurate to arc-seconds. Aim a degree off and the signal, already an attowatt, drops by another 3 dB or worse.

⚠️ Common Misconception: "A higher-gain antenna is simply a better antenna." Gain is not free quality; it is focus, and focus is a trade. A high-gain dish concentrates power into a pencil beam, which is wonderful if you can aim it precisely at Earth and terrible if you cannot. Many spacecraft therefore carry both: a high-gain dish for fast data return when the attitude is known and stable, and a low-gain antenna — nearly isotropic, a stub that radiates in almost all directions — for emergencies, when the craft is tumbling or in safe mode and cannot point. The low-gain antenna's link is agonizingly slow, but it works when you have lost attitude control and need to hear the spacecraft say "I am alive, here is why I rebooted." Choosing the gain is choosing where on the focus-versus- robustness spectrum a given link should sit.

🔗 Connection: The same $G = \eta(\pi D/\lambda)^2$ works for the receiving dish, which is why the Deep Space Network's 70 m antenna reaches $74\ \text{dBi}$ — it is 2000 wavelengths across at X-band. Transmit gain and receive gain enter the Friis equation symmetrically, so a decibel bought by enlarging the ground antenna is exactly as valuable as a decibel bought by enlarging the spacecraft's — and the ground antenna is far easier to enlarge, since it does not have to be launched at the exponential cost of Chapter 3. This asymmetry — cheap mass on the ground, ruinously expensive mass in flight — is why humanity built a few gigantic ears rather than putting gigantic mouths on every probe.

26.3 The Deep Space Network

Those gigantic ears have a name.

Definition (Deep Space Network). The Deep Space Network (DSN) is NASA's globe-spanning array of large ground antennas that communicate with and track interplanetary spacecraft. It comprises three Deep Space Communications Complexes spaced roughly 120° apart in longitude — near Goldstone (California, USA), Madrid (Spain), and Canberra (Australia) — each with one 70 m antenna and several 34 m antennas. Managed by NASA's Jet Propulsion Laboratory, it is the primary link for essentially every mission beyond Earth orbit.

The three-site, 120°-apart arrangement is not aesthetic; it is forced by a rotating Earth. A single antenna can see a given patch of sky for only part of a day before the Earth turns it away. Three complexes, each able to watch about a third of the way around, guarantee that any target near the plane of the planets stays above the horizon of at least one complex at all times. As Goldstone rotates out of view of a probe, Canberra or Madrid is rotating in, and the track is handed off without a gap. Continuous contact with the whole solar system, from three points on one planet — that is the geometric idea.

📜 From History: the dishes that grew for Voyager. The DSN's biggest antennas began life as 64 m dishes in the 1960s, sized for Mariner and Apollo-era distances. By the 1980s a harder job loomed: Voyager 2 would reach Neptune in 1989 at 30 AU, where the signal would be four times weaker than at Uranus. So between 1982 and 1988 NASA rebuilt all three 64 m antennas into 70 m antennas — a modest-sounding change that, because gain scales as diameter squared, added about 1.9 dB, and, combined with arraying (below), let Voyager return sharp images of Neptune and Triton at about 21.6 kbit/s. The infrastructure on the ground was upgraded to meet a spacecraft that had been launched a decade earlier and could not be changed. It is a recurring pattern in deep space: the probe is fixed at launch; the ground evolves to keep hearing it.

Two ideas make the DSN more than just three big dishes. The first is a single figure of merit that folds a receiving station's whole quality into one number: $G/T$, the ratio of antenna gain to system noise temperature, in dB/K. It is exactly the combination that appears in the link budget's $P_r/N_0$, since $P_r/N_0 = \text{EIRP} - L_{\text{fs}} - L_{\text{other}} + (G/T) - 10\log_{10}k$. Raising a station's $G/T$ — a bigger dish (more $G$) or a colder amplifier (less $T$) — improves every link it supports. The DSN 70 m at X-band has $G/T \approx 74 - 10\log_{10}(20) \approx 61\ \text{dB/K}$, among the best on Earth.

The second idea is arraying: electronically combining the signals from several antennas so they act as one larger antenna. Combining the 70 m dish with a few 34 m dishes (or, historically, with the Parkes radio telescope in Australia for Voyager) adds their collecting areas, raising $G/T$ by a few decibels — and a few decibels, on the wrong side of the inverse-square wall, is the difference between one image and ten. Arraying is how the DSN musters extra gain for the critical hours of a planetary encounter without building a single impossibly large dish.

🔗 Connection: The DSN is a shared, oversubscribed resource — a few dozen antennas for a hundred-odd active deep-space missions — so time on the network is a currency mission planners fight over, and scheduling it is part of the operations concept you will meet in Chapter 31. A mission that needs more downlink than its DSN allocation allows must either compress its data harder, carry a bigger antenna, or accept returning less science. Communications is not only a physics problem; it is a logistics problem shared across the whole community of deep-space explorers.

🔄 Check Your Understanding 1. Why does the DSN need at least three sites rather than two, and why are they placed ~120° apart in longitude specifically (rather than latitude)? 2. Arraying a 70 m dish with three 34 m dishes: roughly how much does it help? (Compare collecting areas; answer in dB.)

Answers

1. Earth rotates about its polar axis, sweeping antennas through longitude, so a target sits above a given site for only part of each day; three sites ~120° apart in longitude keep any near-ecliptic target in view of at least one at all times (two sites 180° apart would leave gaps near the horizon and no overlap for handover). Latitude spacing would not help with the daily rotation. 2. Collecting area scales as $D^2$: the 70 m has area $\propto 70^2 = 4900$; each 34 m has $\propto 34^2 = 1156$, so three add $3468$, for a total of $8368$ versus $4900$ — a factor of $1.71$, or $10\log_{10}(1.71) = 2.3\ \text{dB}$. Modest in decibels, decisive at Neptune.

26.4 Bands, modulation, coding, and data rates

We have treated frequency as a given. It is in fact a design choice with deep consequences, governed by international allocations and by physics. Deep-space communication lives mainly in three bands:

The physics of the choice is a beautiful near-cancellation, and it is worth getting exactly right. Move to a higher frequency (shorter $\lambda$) and the free-space path loss rises as $f^2$ — the term $(4\pi d/\lambda)^2$ grows when $\lambda$ shrinks. That sounds bad. But for a fixed physical dish at each end, the antenna gain also rises as $f^2$ (recall $G = \eta(\pi D/\lambda)^2$), and it does so at both ends. Track the powers of $f$ through the Friis equation with the gains written in terms of fixed areas $A_t, A_r$:

$$ P_r = P_t\,\frac{4\pi A_t}{\lambda^2}\cdot\frac{4\pi A_r}{\lambda^2}\cdot\left(\frac{\lambda}{4\pi d}\right)^2 = P_t\,\frac{A_t A_r}{\lambda^2 d^2} \;\propto\; \frac{1}{\lambda^2} \;\propto\; f^2. $$

The two gains beat the one path loss, and the received power rises as $f^2$. For fixed dish sizes, higher frequency wins — which is why the field marched from S-band to X-band and is now moving to Ka-band. The price is the narrower beam (harder pointing, back to Section 26.2) and greater atmospheric loss: Ka-band is absorbed by water vapor and rain, so a Ka downlink can be rained out, forcing missions to keep an X-band fallback or to schedule around weather at the ground stations.

🔧 Engineering Reality: The $f^2$ win assumes you are aperture-limited — that both dishes are as big as they can be and you are squeezing more out of them. That is exactly the deep-space situation. But the narrower Ka beam demands better pointing at both ends, and the DSN's Ka rain losses mean the gain is statistical: a mission designs to a link that closes, say, 90% of the time, and accepts occasional weather outages. Engineering the band choice means balancing the $f^2$ data-rate prize against pointing hardware, weather margin, and the maturity of the flight and ground equipment.

Once the signal arrives, two more layers decide how much information it carries. Modulation is how bits are impressed on the carrier; deep space favors phase-shift keying (BPSK, QPSK), which packs bits into the phase of the wave and is robust at the low signal-to-noise ratios that are all deep space can offer. Channel coding — forward error correction — is the quiet hero of the story.

💡 Intuition: You cannot ask Pluto to "say that again." A retransmission request would take nine hours to round-trip, and the encounter would be over. So deep-space links send extra, carefully structured redundant bits — an error-correcting code — that let the receiver reconstruct the original data even when many bits arrive corrupted, with no second copy needed. The redundancy is not waste; it is what makes the whisper legible.

The payoff of coding is measured as coding gain: the reduction in the $E_b/N_0$ a receiver needs to hit a target error rate. Uncoded phase-shift keying needs about $9.6\ \text{dB}$ of $E_b/N_0$ for a bit error rate of $10^{-5}$. The concatenated codes Voyager used (a convolutional code wrapped in a Reed-Solomon code) cut that to roughly $2.5\ \text{dB}$ — a 7 dB coding gain, which through the link budget is worth a factor of five in data rate or, equivalently, letting the probe travel more than twice as far. Modern turbo codes and LDPC codes push the required $E_b/N_0$ down toward $1\ \text{dB}$, approaching the ultimate Shannon limit of $-1.6\ \text{dB}$ below which no code of any kind can work. Every decibel of coding gain is a decibel you did not have to buy with a bigger dish or a closer approach — which is why the epigraph's Shannon presides over this chapter.

Worked Example: New Horizons at Jupiter versus Pluto — the inverse-square in data. New Horizons flew past Jupiter (≈ 4.9 AU) in 2007 and Pluto (≈ 32.3 AU) in 2015 with the same transmitter, dish, and ground stations. Only the distance changed. Since received power — and therefore data rate — falls as $1/d^2$, the rate ratio is $$\frac{R_{\text{Jupiter}}}{R_{\text{Pluto}}} = \left(\frac{d_{\text{Pluto}}}{d_{\text{Jupiter}}}\right)^2 = \left(\frac{32.3}{4.9}\right)^2 = (6.59)^2 \approx 43.$$ If Pluto returns ~1 kbit/s, the identical system at Jupiter returns ~43 kbit/s — and indeed New Horizons downlinked far faster at Jupiter. The mission did not get worse; it got farther, and the inverse-square law did the rest. This is the same tyranny, now measured in bits instead of watts.

🔄 Check Your Understanding 1. Two missions use the same dishes at both ends. One works at X-band (8.4 GHz), one at Ka-band (32 GHz). By how many dB is the Ka-band link stronger, ignoring weather? (Use $P_r \propto f^2$.) 2. A coding scheme lowers the required $E_b/N_0$ from 9.6 dB to 2.6 dB. By what factor can the data rate increase, all else equal?

Answers

1. $P_r \propto f^2$, so the gain is $20\log_{10}(32/8.4) = 20\log_{10}(3.81) = 11.6\ \text{dB}$ — before subtracting Ka's larger weather losses, which is why Ka's real-world advantage is smaller. 2. The 7 dB of coding gain becomes 7 dB of extra $P_r/N_0$ available for rate: $10^{7/10} = 5.0$, so about a fivefold increase.

26.5 Navigation: ranging, Doppler, and delta-DOR

The radio link is not only a data pipe; it is also a measuring instrument. The same carrier that brings the pictures home tells us, with astonishing precision, where the spacecraft is and how fast it is moving — the raw observations that orbit determination (Chapter 13) turns into a trajectory, and that guidance, navigation, and control (Chapter 27) acts on. Three techniques, together, pin down all the coordinates.

Ranging: how far

Definition (ranging). Ranging measures the distance to a spacecraft by timing a signal's round trip. A ground station transmits a modulated signal; the spacecraft's transponder coherently retransmits it; the station measures the round-trip light time $\tau$ and computes the distance as $d = c\,\tau/2$.

Because $c$ is known to exquisite precision, distance becomes a timing measurement, and radio timing is superb. Timing the round trip to a nanosecond fixes the one-way distance to about $c\times(10^{-9})/2 \approx 0.15\ \text{m}$; in practice the DSN ranges to a few meters over interplanetary distances, after carefully removing the delays added by the charged solar-system plasma and Earth's atmosphere.

Worked Example: the round trip to Pluto. With New Horizons at $d = 4.83\times10^{12}\ \text{m}$, the round-trip light time is $$\tau = \frac{2d}{c} = \frac{2(4.83\times10^{12})}{2.998\times10^8} = 3.22\times10^4\ \text{s} \approx 8.95\ \text{hours}.$$ A command sent to New Horizons and its acknowledgment are separated by about nine hours. There is no joystick across the solar system: every risky moment — the Pluto flyby sequence, an engine burn, a reboot — must be handled by the onboard computer (Section 26.6) and by pre-loaded command sequences, because by the time Earth could react, it would be far too late. This nine-hour gulf is the operational face of the light-speed limit, and it drives the push toward autonomy.

Doppler tracking: how fast (toward or away)

Definition (Doppler tracking). Doppler tracking measures a spacecraft's line-of-sight velocity from the frequency shift of its carrier. A target receding at radial speed $v$ shifts the received frequency by $\Delta f / f \approx -v/c$ (for $v \ll c$); measuring $\Delta f$ yields $v$. Using a coherent transponder, the ground compares the returned frequency to the transmitted one for a precise two-way measurement.

Doppler is the complement of ranging: ranging gives distance, Doppler gives the rate of change of distance. And it is fantastically sensitive.

Worked Example: Doppler shift and its precision. A station transmits at $f = 7.15\ \text{GHz}$ (X-band uplink) to a probe receding at $v = 10\ \text{km/s}$. The one-way Doppler shift is $$\Delta f = -\frac{v}{c}\,f = -\frac{10^4}{2.998\times10^8}(7.15\times10^9) = -2.4\times10^5\ \text{Hz} = -238\ \text{kHz}.$$ The two-way shift is about twice this. Now the precision: the DSN can measure the carrier frequency to a small fraction of a hertz. Measuring $\Delta f$ to $0.001\ \text{Hz}$ pins the velocity to $$\delta v = \frac{c\,\delta f}{f} = \frac{(2.998\times10^8)(0.001)}{7.15\times10^9} \approx 4\times10^{-5}\ \text{m/s} = 0.04\ \text{mm/s}.$$ A spacecraft's radial velocity, across billions of kilometers, known to a few hundredths of a millimeter per second. That is not a typo. It is why Doppler tracking can detect the faint tug of an unseen moon on a flyby, or the whisper of thrust from a leaking valve.

Delta-DOR: where on the sky

Ranging and Doppler nail the distance and the range-rate — the coordinate along the line of sight. They say almost nothing about the transverse position: whether the probe is a little to the left or right, up or down, as seen from Earth. That plane-of-sky coordinate is what delta-DOR supplies.

Definition (delta-DOR). Delta-DOR (Delta Differential One-way Ranging) measures a spacecraft's angular position on the sky using two widely separated ground antennas as an interferometer. The stations simultaneously record the spacecraft's signal; the difference in signal arrival time between them fixes one plane-of-sky angle. To cancel common errors (clocks, plasma, atmosphere), they then observe a quasar of precisely known celestial position near the same line of sight, and take the difference of the two measurements — the "delta."

The geometry is that of any two-element interferometer. Two stations separated by a baseline $B$ (up to nearly an Earth diameter, ~8,000–10,000 km between complexes) see the same distant signal arrive with a time difference $\Delta t = (B/c)\sin\theta$, where $\theta$ is the angle of the source from the baseline's perpendicular. Measure $\Delta t$, invert for $\theta$, and you have an angle. Differencing against the quasar — a natural radio beacon whose direction is known to sub-nanoradian accuracy from decades of radio astronomy — subtracts off the errors common to both observations, leaving the spacecraft's angle relative to a fixed point on the sky.

Worked Example: delta-DOR accuracy at Mars. Delta-DOR achieves angular accuracies of a few nanoradians. At Mars, a typical range is $d \approx 1.5\ \text{AU} = 2.25\times10^{11}\ \text{m}$, so an angular error of $5\ \text{nrad}$ corresponds to a plane-of-sky position error of $$\delta x = d\,\delta\theta = (2.25\times10^{11})(5\times10^{-9}) \approx 1{,}100\ \text{m}.$$ About a kilometer, across the line of sight, at Mars — the missing coordinate that ranging and Doppler cannot provide, essential for delivering a lander to its target ellipse. Achieving it requires timing the two stations' recordings to tens of picoseconds: to reach $1\ \text{nrad}$ on an $8{,}000\ \text{km}$ baseline needs $\delta(\Delta t) = (B/c)\,\delta\theta \approx 27\ \text{ps}$, which is exactly why the quasar calibration is indispensable.

🔗 Connection: These three observables — range, range-rate, and plane-of-sky angle — are the input to orbit determination (Chapter 13), which statistically fuses many of them over time into a best-estimate trajectory and its uncertainty. That trajectory, in turn, is what the guidance system of Chapter 27 uses to decide when and how to burn. Communications, navigation, and control are one continuous loop: the radio measures the state, the ground computes the orbit, the plan updates, and commands flow back up the very same link. You cannot separate talking to a spacecraft from knowing where it is.

🔄 Check Your Understanding 1. Which of ranging, Doppler, and delta-DOR gives you the coordinate across the sky, and why can't ranging or Doppler provide it? 2. A probe is 3 AU away. Estimate the round-trip light time. What does it imply for "flying" the spacecraft in real time?

Answers

1. Delta-DOR. Ranging measures distance and Doppler measures its rate of change — both are purely along the line of sight, so neither senses a small sideways displacement; delta-DOR measures the angle on the sky directly via interferometry. 2. $3\ \text{AU} = 4.5\times10^{11}\ \text{m}$; round trip $= 2d/c = 9\times10^{11}/3\times10^8 = 3{,}000\ \text{s} = 50\ \text{minutes}$. Real-time joystick control is impossible; the spacecraft must act autonomously and follow pre-loaded sequences.

26.6 Onboard computers: radiation-hardened

Everything so far assumed something on the spacecraft decided what to transmit, executed the flyby sequence, ran the attitude loop, and survived nine hours out of contact. That something is the onboard computer — and it is, deliberately, a museum piece.

Definition (radiation-hardened processor). A radiation-hardened ("rad-hard") processor is one designed and manufactured to keep operating correctly in the space radiation environment — energetic protons, heavy cosmic-ray ions, and accumulated gamma dose — through special fabrication, redundant circuit design, and shielding. The cost of that survivability is performance: rad-hard processors typically lag commercial chips by 15–20 years in speed.

Space is full of ionizing radiation with no atmosphere or magnetic field to stop it (a theme from Chapter 1), and it attacks electronics in three ways:

• Single-event upset (SEU): a single energetic particle deposits enough charge in a memory cell or logic node to flip a bit — a soft error. Non-destructive, but it can corrupt data, or, worse, corrupt the program the computer is executing.
• Single-event latchup (SEL): a particle triggers a parasitic short circuit that can destroy the chip unless power is cycled quickly. A hard, sometimes fatal error.
• Total ionizing dose (TID): the cumulative radiation over years slowly degrades every transistor, measured in rad or gray ($1\ \text{Gy} = 100\ \text{rad}$). A part rated for a few krad might fail early in a harsh orbit; rad-hard parts tolerate hundreds of krad to a megarad.

🔧 Engineering Reality: why rad-hard means slow. Two of the main defenses directly cost speed. First, rad-hard chips use larger transistor feature sizes — a bigger node holds more charge, so a passing ion is less likely to flip it — but larger features mean slower switching and lower density. Second, they lean on redundancy: triple-modular redundancy runs three copies of the logic and votes, and error-detecting-and-correcting (EDAC) memory continuously scrubs bit-flips out of RAM. Both trade raw throughput for correctness. The canonical example, the RAD750, flies on Curiosity, many orbiters, and beyond: it runs at roughly 200 MHz (a few hundred MIPS), tolerates several hundred krad, and costs on the order of $200{,}000 — while being descended from a consumer PowerPC chip from 1997. The phone in your pocket is tens of thousands of times faster and costs a few hundred dollars. In space you do not buy speed; you buy the near-certainty that the computer will still be computing after a decade in the radiation bath.

This is theme #2 of the whole book — space is unforgiving — written in silicon. On the ground, a cosmic-ray bit-flip is a once-in-a-blue-moon curiosity; in interplanetary space, an unshielded chip of a square centimeter might suffer an upset every few days, so the system is built assuming errors will happen and must be caught. Watchdog timers reboot a hung processor; EDAC corrects memory as it is read; redundant computers vote or fail over; and a safe mode parks the spacecraft in a stable, Sun-pointed, Earth-listening configuration when anything looks wrong, buying the nine hours it takes for the ground to notice and respond.

🔗 Connection: Redundancy here is the same discipline you will study system-wide in Chapter 32: you cannot repair a spacecraft, so you design so that no single event — a flipped bit, a stuck transistor — ends the mission. The rad-hard computer is where "everything must work, and there is no repair truck" meets the microchip. It is slow because it must be sure.

🐛 Find the Error. A systems engineer proposes flying the latest commercial mobile processor to a Jupiter orbiter, reasoning: "It is 30,000 times faster than a RAD750, so even if we have to reboot it ten times a day from radiation upsets, we come out enormously ahead on compute." What is wrong with the reasoning?

Answer

Two things. First, single-event latchup at Jupiter — the harshest radiation environment in the solar system — can destroy a commercial chip, not merely reboot it, so "reboot ten times a day" understates the risk: the part may not survive at all. Second, and more subtly, a reboot is not free: while the computer is down or recovering, the spacecraft is not running its attitude-control loop, not pointing its antenna, and not executing its sequence — and with a nine-hour (Pluto) or ~100-minute (Jupiter) round-trip light time, the ground cannot step in. Ten outages a day could mean losing attitude, tumbling off Earth-point, and dropping the link entirely. Raw compute is worthless if the platform it runs on cannot keep the spacecraft alive; the whole point of rad-hard is availability, not speed. The right architecture is often a rad-hard processor for the critical control functions, with commercial parts, if used at all, quarantined to non-critical, restartable tasks behind the rad-hard watchdog.

Mission Design Checkpoint: your comms note and astrotools/comms.py

Every chapter of Part IV adds a subsystem to your Mission Design Review. This one adds the communications note — and the astrotools module that lets you size it.

The design. Open your MDR and add a communications subsystem note answering four questions for your chosen mission (Track A GEO comsat, B lunar lander, C Mars orbiter, or D asteroid rendezvous):

1. Band. S, X, or Ka? Near Earth (Tracks A, B) you have more link margin and can use higher bands for high data rates; deep space (Tracks C, D) usually means X-band, with Ka where the data volume and pointing justify it.
2. Data rate and volume. How many bits per second must you return, and how much per day? An imaging Mars orbiter may need tens of Mbit/s; a distant asteroid probe may live on a few kbit/s.
3. Ground stations. Near-Earth commercial ground networks for Tracks A/B; the Deep Space Network for Tracks C/D — and remember DSN time is scarce and scheduled.
4. Antenna and power. A high-gain dish sized to close the link, plus a low-gain antenna for safe mode; and the transmitter power drawn from your power budget — often the largest single load.

The code. Create astrotools/comms.py, the link-budget module, working entirely in decibels:

import math

C = 2.998e8               # speed of light, m/s
K_BOLTZMANN = 1.380649e-23  # J/K

def watts_to_dbw(p_w):
    """Power in watts -> dBW."""
    return 10 * math.log10(p_w)

def fspl_db(dist_m, freq_hz):
    """Free-space path loss (dB) over dist_m at freq_hz. Note the 20*log10."""
    wavelength = C / freq_hz
    return 20 * math.log10(4 * math.pi * dist_m / wavelength)

def dish_gain_db(diameter_m, freq_hz, efficiency=0.6):
    """Parabolic-dish gain (dBi) from diameter and frequency."""
    wavelength = C / freq_hz
    return 10 * math.log10(efficiency * (math.pi * diameter_m / wavelength) ** 2)

def link_budget(pt_w, gt_db, gr_db, freq_hz, dist_m, extra_loss_db=0.0):
    """Received power (dBW) for a one-way link. Adds dB, so it just sums the column."""
    eirp = watts_to_dbw(pt_w) + gt_db
    return eirp - fspl_db(dist_m, freq_hz) + gr_db - extra_loss_db

# New Horizons at Pluto (Section 26.1), hand-traced:
gt = dish_gain_db(2.1, 8.4e9, 0.6)
gr = dish_gain_db(70.0, 8.4e9, 0.7)
pr = link_budget(12, gt, gr, 8.4e9, 4.83e12, extra_loss_db=2.0)
print(f"EIRP = {watts_to_dbw(12) + gt:.1f} dBW")
print(f"free-space path loss = {fspl_db(4.83e12, 8.4e9):.1f} dB")
print(f"received power = {pr:.1f} dBW ({10**(pr/10):.1e} W)")
# Expected output:
# EIRP = 53.9 dBW
# free-space path loss = 304.6 dB
# received power = -178.5 dBW (1.4e-18 W)

The full worked increment — a GEO comsat downlink for Track A, computed with these same functions — is in code/project-checkpoint.py. In later chapters comms.py joins the rest of astrotools: the power draw of the transmitter feeds the power budget, and the data rate you compute here sets how much of your mission's science actually comes home. A spacecraft is only as good as its downlink.

Summary

Communications, navigation, and data handling are the thread that makes a spacecraft useful — and the inverse-square law is the adversary that runs through all of them.

Numbers worth remembering: New Horizons at Pluto received ~$1.4\times10^{-18}\ \text{W}$ at ~1 kbit/s; round-trip light time ~9 hours. Every factor of 10 in power is 10 dB; every factor of 2 is ~3 dB. Received power and data rate both fall as $1/d^2$.

Spaced Review

Retrieval strengthens memory. Answer from memory before checking, then look back at the cited chapter. This set revisits Chapter 25 — Power Systems, because the transmitter you just sized is usually the biggest customer of the power you learned to generate there.

1. (Ch. 25) New Horizons cannot use a solar array. In one sentence, why, and what does it use instead to generate its ~200 W?
2. (Ch. 25) If a spacecraft's transmitter needs 12 W of radio-frequency output but the amplifier is only ~40% efficient, roughly how much electrical power does it draw from the bus — and why does that make the comms subsystem a headline item on the power budget?
3. (Ch. 25) During an eclipse a spacecraft runs on batteries. Why does the depth of discharge you allow limit how long, and how often, you can run a power-hungry transmitter in shadow?
4. (Ch. 25) A solar array degrades over a mission's life. If your end-of-life power is 30% below beginning-of-life, how should that change the transmitter power (and therefore data rate) you promise in year ten versus year one?

Answers

1. At ~32 AU the sunlight is about $1/32^2 \approx 0.1\%$ of Earth's, far too feeble for a practical solar array, so New Horizons carries a radioisotope thermoelectric generator (RTG) that turns the heat of decaying plutonium-238 into electricity. 2. About $12/0.40 = 30\ \text{W}$ of electrical power (the other ~18 W becomes waste heat the thermal system must reject); on a spacecraft with only a couple hundred watts total, a 30 W transmitter is a large, and often the largest, single load — which is why data rate is ultimately limited by power as much as by distance. 3. Batteries age faster the more deeply you discharge them each cycle, so a strict depth-of-discharge limit caps the energy available per eclipse; a transmitter that draws tens of watts can exhaust that budget quickly, forcing you to transmit less (or not at all) in shadow. 4. You must size the link to the end-of-life power: the honest data rate you can promise in year ten is set by the degraded array, so either the link is designed conservatively from the start or the data rate is planned to decline as the power does.

What's Next

We now have a spacecraft that can generate power, survive the cold, hold its shape, tell Earth what it sees, and know where it is to the meter and the millimeter-per-second. What it still cannot do is decide, moment to moment, what to do about it — how to point, when to burn, how to fly itself through the nine hours when no one on Earth can help. Navigation gave us the estimate of the state; now we must close the loop. In Chapter 27 we assemble guidance, navigation, and control: we take the orbit determination of Chapter 13 and the attitude dynamics of Chapter 14, add the navigation observables of this chapter, and build the feedback loop — sensors to estimation to actuation — that actually flies the vehicle, from a rocket's ascent to the "seven minutes of terror" of a Mars landing that no ground controller can touch.