Modulation Schemes for Satcom in 2026: QPSK, 8PSK, 16APSK, and 32APSK Explained

Why Modulation Is the Beating Heart of a Satcom Link

Every satellite link is a negotiation between three merciless constraints: how much power the spacecraft can pump into its high-power amplifier, how much bandwidth the regulator will let you occupy, and how much noise the physics of a 36,000 km path (for geostationary orbit) drops into your receiver. Modulation is the mathematical choice you make to balance those three. Get it right and a 72 MHz Ku-band transponder pushes 250+ Mbps of usable throughput to a 1.2 m VSAT dish during a rainstorm. Get it wrong and the same transponder can barely close the link on a clear day.

In 2026, the practical satcom modulation menu is short but nuanced: QPSK, 8PSK, 16APSK, and 32APSK dominate DVB-S2 and DVB-S2X carriers, with QPSK and its OQPSK / pi/4-QPSK variants also carrying most of the narrowband voice, telemetry, and command traffic. Higher-order QAM (64QAM, 256QAM), which is standard in terrestrial fiber and 5G, is almost never used for satellite downlinks. The reason is not academic. It is the traveling-wave tube amplifier (TWTA), or in newer platforms the gallium-nitride solid-state power amplifier (GaN SSPA), and the way those devices distort signals that have a large peak-to-average power ratio (PAPR).

This article walks through each scheme the way we teach it in the Refonte Learning satellite communications engineering program: the constellation geometry, the required Es/N0, the coding companion, the operational quirks, and where each one earns its slot on a modern MODCOD table. If you have ever stared at a DVB-S2X ACM table and wondered why 16APSK 3/4 is not simply better than 8PSK 5/6, or why nobody puts 256QAM through a saturated TWTA, this is for you. We will also address the People Also Ask questions that follow satcom engineers around: what modulation does satcom actually use, what the four canonical modulation families are, and why 256QAM is such an awkward fit for a satellite downlink.

The Four Canonical Modulation Families (and Where Satcom Fits)

When someone asks 'what are the four types of modulation', they are usually pointing at the textbook taxonomy that groups every practical scheme under one of four umbrellas: amplitude modulation (AM), frequency modulation (FM), phase modulation (PM), and their digital descendants. In digital communications, that same taxonomy collapses to amplitude-shift keying (ASK), frequency-shift keying (FSK), phase-shift keying (PSK), and quadrature amplitude modulation (QAM), with amplitude-phase-shift keying (APSK) sitting as a hybrid.

Satcom uses almost exclusively the phase and phase+amplitude families. Pure ASK is dead in modern satellite links because amplitude-only signals waste power and have terrible spectral properties. FSK survives in a few niches (some low-rate telemetry, S-band command uplinks, older Inmarsat air-interface variants, and LoRa-style non-satellite systems) but is not competitive on spectral efficiency for broadband payloads. That leaves PSK and APSK as the workhorses.

A good mental model:

• BPSK: 1 bit per symbol, most robust, used for spacecraft telemetry and command (TT&C), acquisition preambles, and deep-space links where every dB of margin matters.
• QPSK: 2 bits per symbol, the baseline for DVB-S, most VSAT return links, and the fallback MODCOD when rain fade hits.
• 8PSK: 3 bits per symbol, the DVB-S2 sweet spot for clear-sky Ku-band forward links.
• 16APSK / 32APSK: 4 and 5 bits per symbol, DVB-S2 and DVB-S2X high-efficiency modes for well-linearized amplifiers.
• 64APSK / 128APSK / 256APSK: added by DVB-S2X for professional feeder links and DTH contribution circuits with very high C/N.
• QAM: rare in satcom because the square QAM constellation punishes non-linear amplifiers, though it appears in some intra-fiber gateway hops and in newer 5G non-terrestrial network (NTN) waveforms where the amplifier is well behaved.

The entire satcom modulation story is therefore not really about picking between AM, FM, PM, and QAM. It is about picking between constellations that have been specifically shaped to survive a non-linear amplifier and a noisy, long-delay channel. Everything below flows from that constraint.

QPSK: The Baseline That Refuses to Die

QPSK carries two bits per symbol by placing four constellation points on a unit circle at 45, 135, 225, and 315 degrees. Because all four points sit on the same amplitude ring, the signal envelope is nominally constant (before pulse shaping), which means a TWTA driven near saturation barely distorts it. That single geometric property is why QPSK has never gone away.

On the DVB-S2 MODCOD table, QPSK is paired with LDPC code rates from 1/4 up through 9/10. At rate 1/2, a QPSK carrier needs about 1.0 dB Es/N0 for quasi-error-free operation (packet error rate below 1e-7 at the LDPC output, and effectively zero at the outer BCH output). At rate 9/10, the requirement climbs to roughly 6.4 dB. That range, roughly 1 to 6.5 dB Es/N0, is the working envelope where QPSK earns its keep.

Operationally, QPSK does three jobs:

1. Rain-fade fallback in ACM systems: When a Ka-band gateway sees a rain cell over a remote terminal, the modulator can drop that terminal from 16APSK 3/4 down to QPSK 2/3 within a superframe. The terminal keeps the session alive at reduced throughput instead of losing sync entirely.
2. Return-channel modulation on VSATs: The remote terminal's SSPA is small (2 to 8 W typical) and cheap, and its antenna gain is modest. QPSK gives enough margin to close the link at low symbol rates without demanding a linear amplifier.
3. TT&C and safety-of-life: Telemetry, tracking, and command channels almost always use BPSK or QPSK because they must close under worst-case attitude, rain, and antenna-off-boresight conditions.

QPSK also has two important variants worth knowing. Offset QPSK (OQPSK) staggers the in-phase and quadrature bit transitions by half a symbol so the trajectory never crosses the origin, cutting PAPR and easing amplifier linearity requirements further. It is common on TT&C links and on some older Iridium and GPS signals. Pi/4-QPSK rotates the constellation by 45 degrees every symbol, which similarly reduces envelope excursions and appears in TETRA and some legacy mobile-satellite standards. In DVB-S2X, a specialized 'linear QPSK' variant is defined for very low SNR (VL-SNR) frames, extending operation down to roughly minus 10 dB Es/N0 with heavy coding, which matters for mobile satellite terminals on ships and aircraft.

8PSK: The Clear-Sky Workhorse

8PSK adds four more phase states, giving eight points equally spaced around a single amplitude circle and carrying three bits per symbol. On paper, 8PSK is a 50 percent throughput uplift over QPSK for the same bandwidth. In practice, the trade is steeper than it looks.

Moving from QPSK to 8PSK compresses the angular distance between adjacent constellation points from 90 degrees to 45 degrees. Noise, phase jitter, and amplifier AM-to-PM conversion all attack that angular distance directly. The Es/N0 required for 8PSK at LDPC rate 3/4 is roughly 7.9 dB, compared with about 4.0 dB for QPSK 3/4. You are paying almost 4 dB of link budget for a 1.5x throughput gain. That is a good trade only when you have the margin, meaning a clear-sky link, a large receive antenna, or a benign frequency band.

8PSK is the natural home for:

• DTH direct-to-home broadcast in Ku-band at high C/N receive sites (larger dishes, dry climates)
• Broadband forward links to fixed VSATs during clear-sky conditions
• Contribution and backhaul feeds where the ground segment can throw a 4.5 m or 7.3 m antenna at the problem

Because 8PSK is still a constant-envelope scheme (all points on one amplitude ring), a well-tuned TWTA can operate closer to saturation than it can with an APSK signal, though not as close as with QPSK. Typical operating back-off (OBO) for 8PSK on a legacy TWTA is around 1.5 to 2 dB, versus 0.5 to 1 dB for QPSK. That amplifier back-off comes straight out of your EIRP, and it is one of the underappreciated costs of higher-order modulation.

Understanding those back-off numbers is one of the things we drill in class when covering the satellite subsystems side of the link. The payload's power amplifier, its channel filter, and the modulator are not independent choices; they co-determine which MODCODs are actually reachable in service.

16APSK: When Amplitude Enters the Chat

16APSK is where the modulation world stops being purely phase-based and starts admitting amplitude information. The DVB-S2 16APSK constellation places 4 points on an inner ring and 12 points on an outer ring, with a specific radius ratio (gamma) chosen to maximize the minimum Euclidean distance between points after non-linear amplification. Typical gamma values are 2.57 to 3.15 depending on code rate.

Why two rings instead of a square 16QAM grid? Because a square 16QAM has three distinct amplitude levels (with corner points at the highest amplitude), and the peak-to-average excursions drive a TWTA hard into its non-linear region. The two-ring APSK design keeps the number of amplitude levels down to two, which is much easier for the amplifier's pre-distorter or for careful OBO management. DVB-S2X extends the family with alternative 16APSK constellations (labeled 16APSK-L for 'linear channel') that are geometrically closer to 16QAM and can be used on well-linearized ground-station transmitters or when the space segment uses a linearized channel amplifier (LCAMP).

Required Es/N0 for 16APSK at LDPC rate 3/4 sits near 10.2 dB, and at rate 5/6 near 11.6 dB. Spectral efficiency reaches roughly 3.0 to 3.3 bits per second per Hz of allocated symbol rate, before roll-off factor overhead. Compared with 8PSK 5/6 (2.5 bit/s/Hz at ~9.4 dB), 16APSK 3/4 gives you a ~20 percent throughput gain for roughly 1 dB more required Es/N0. In a clear-sky Ka-band link with a 1.2 m antenna and a modern gateway, that trade is usually worth it.

Operational reality check: 16APSK forces the payload TWTA into deeper back-off, typically 2.5 to 4 dB depending on the pre-distorter. That is real EIRP you are giving up. On a bent-pipe transponder shared across multiple carriers, aggregating multiple 16APSK carriers into one TWTA also raises intermodulation concerns, so the operator has to run link budgets carrier by carrier and check the composite spectrum for spectral regrowth into adjacent channels. This is exactly the kind of interaction our students see when they work through the ground segment and space segment hand-off exercises.

32APSK and Beyond: Squeezing the Last Bits

32APSK adds a third amplitude ring, packing 32 points across four inner points, twelve middle, and sixteen outer (in the DVB-S2 layout). Five bits per symbol, spectral efficiency in the 3.7 to 4.4 bit/s/Hz range depending on code rate, and Es/N0 requirements starting around 11.0 dB at rate 3/4 and reaching about 13.1 dB at rate 9/10.

DVB-S2X extends this further with 64APSK, 128APSK, and 256APSK, along with a rich set of intermediate MODCODs designed to fit specific link margins. 64APSK, for example, sits around 14 to 16 dB required Es/N0 and pushes past 5 bit/s/Hz. Nobody uses 128APSK or 256APSK on a bent-pipe TWTA in saturation. They exist for professional feeder links, gateway-to-gateway hops via satellite, DTH contribution networks, and increasingly for regenerative payloads that can transmit on ground-station-quality linear amplifiers.

The critical thing to internalize about the APSK family is that the constellation is not a fixed drawing. DVB-S2X gives you multiple 16APSK, 32APSK, and 64APSK constellations tuned for different channel linearities, and the modulator picks the right one based on the target link. That is one of the reasons DVB-S2X added roughly 20 new MODCODs on top of DVB-S2: the goal was finer granularity so ACM could match the actual link condition within 0.5 to 1 dB rather than jumping in 2 to 3 dB steps.

The practical failure mode with 32APSK and above is spectral regrowth. When you drive an insufficiently linearized amplifier with a high-PAPR signal, the sidelobes of the output spectrum rise, and you violate the spectral mask that your regulator has assigned to your slot. That can mean interference into adjacent transponders and, in a shared-satellite context, an angry phone call from the operator. Verifying the output spectrum against the mask is part of routine satellite integration and testing, and it is one of the reasons pre-distortion is now standard on any modulator selling into the DVB-S2X market.

LDPC and BCH: The Coding Companions That Make It All Work

Every MODCOD in DVB-S2 and DVB-S2X is a pair: a modulation scheme plus a specific concatenated forward error correction (FEC) scheme. The inner code is a low-density parity-check (LDPC) code with block lengths of 64,800 bits (normal frame) or 16,200 bits (short frame). The outer code is a Bose-Chaudhuri-Hocquenghem (BCH) code that cleans up the residual errors LDPC leaves behind.

LDPC codes are what make modern satcom feasible at the SNRs quoted above. Without them, QPSK 1/2 would need something like 5 to 6 dB more Es/N0 to hit the same error rate. The LDPC codes used in DVB-S2 and DVB-S2X are within roughly 0.6 to 1.0 dB of the Shannon capacity limit for their respective code rates, which is remarkable engineering.

Code rate choice interacts with modulation choice in ways that are easy to get wrong. Some rules of thumb:

• Higher-order modulation (16APSK, 32APSK) usually pairs better with high code rates (3/4, 5/6, 9/10) because the modulation already carries most of the throughput and heavy coding would waste it.
• Lower-order modulation (QPSK) pairs well with low code rates (1/4, 1/3, 2/5) for extreme fade conditions where you need every dB of coding gain.
• The 'sweet spot' for most commercial forward links is somewhere in the 8PSK 3/4 to 16APSK 3/4 range, where both spectral efficiency and coding gain are near-optimal.

BCH is a much simpler code with a small overhead (192 bits for the long frame, 168 for the short) and low error-correction capability, but its job is to reduce the error floor of LDPC from something like 1e-7 to something below 1e-11. That is what lets DVB-S2 quote quasi-error-free performance as a formal spec.

DVB-S2X also adds low-density generator-matrix (LDGM) style extensions in some experimental profiles, though these are less common in deployed hardware. The main DVB-S2X additions are new MODCODs, VL-SNR frames for mobile terminals, and channel bonding (aggregating multiple physical carriers into one logical stream for very high-rate services like 4K and 8K UHD contribution).

Why 256QAM Almost Never Shows Up on a Satellite Downlink

'What is 256QAM modulation' is one of the questions that follows terrestrial-trained engineers into satcom classes, and the honest answer is that 256QAM is beautiful and useful and largely irrelevant to bent-pipe satellite downlinks. Here is why.

256QAM packs 256 constellation points on a 16x16 square grid, giving 8 bits per symbol. It is standard in DOCSIS cable modems, some 5G downlinks, terrestrial microwave backhaul, and Wi-Fi 6/7. In fiber and copper, where amplifiers are linear and SNR is high, 256QAM is a workhorse.

On a satellite downlink, three things go wrong:

1. TWTA non-linearity: The square-QAM constellation has corner points at roughly 1.4x the amplitude of edge points and 2x the amplitude of the center points. That is a PAPR of about 3.7 dB before roll-off, which climbs to 6 to 8 dB after pulse shaping. A TWTA operated at saturation compresses those peaks brutally, warping the corner points inward and destroying the constellation. You would need to back off the TWTA by 6 to 8 dB just to keep the shape, and that back-off is EIRP you no longer have.
2. Required Es/N0: 256QAM at a useful code rate needs something in the 20 to 24 dB Es/N0 range. That is achievable on a gateway feeder link with a 9 m antenna and a clear-sky Ka-band or Q/V-band path, but it is fantasy for a 60 cm consumer VSAT.
3. Standardization inertia: DVB-S2X did not add square QAM to its MODCOD tables. It added higher-order APSK instead (64APSK, 128APSK, 256APSK), specifically because APSK survives amplifier non-linearity better than QAM at the same spectral efficiency.

There are edge cases. Some 5G non-terrestrial network (NTN) proposals use 64QAM or 256QAM on satellite links where the payload is regenerative (demodulate, decode, re-encode, re-modulate on board) and the on-board amplifier is a well-linearized GaN SSPA. In those cases, the space segment behaves more like a repeater in the sky than a bent pipe, and QAM becomes viable. But on classical DVB-S2X commercial satellites, you will see APSK, not QAM, above 8PSK.

Constellation Diagrams in Words: What You Should Picture

Since you cannot see a diagram in this article, here is how to visualize each constellation clearly enough to reason about it.

QPSK: Four dots on a circle at compass positions NE, NW, SW, SE. Equal spacing, equal amplitude. Every symbol is 90 degrees from its neighbors. Grey-coded so that adjacent points differ by only one bit, which minimizes the number of bit errors per symbol error.

8PSK: Eight dots on the same circle, at 22.5, 67.5, 112.5, and so on. Adjacent points are 45 degrees apart. Same amplitude for all. Any phase noise or amplifier AM-to-PM greater than about 15 degrees is going to cause errors.

16APSK: Two concentric circles. Four dots on the inner ring (positions like QPSK but smaller radius) and twelve dots on the outer ring, evenly spaced. The radius ratio (outer/inner) is chosen per code rate, typically 2.57 to 3.15. The bit mapping puts the two amplitude-defining bits first and the three angular bits second.

32APSK: Three rings. Four inner dots, twelve middle, sixteen outer. Radius ratios again optimized per code rate. Angular spacing on the outer ring is 22.5 degrees, so the phase-noise sensitivity is similar to 8PSK, but you also need amplitude discrimination across three levels.

64APSK and above (DVB-S2X): Four or more rings, with progressively denser angular packing on outer rings. At this point, drawing the diagram is more useful as an art piece than as an operational aid; the modulator vendor tunes the exact positions.

One useful reasoning trick: the minimum Euclidean distance between constellation points is what governs symbol error rate at high SNR. When you compare two MODCODs, you can approximate their relative SNR requirements by comparing their minimum distances (normalized to the same average symbol energy). This is why APSK, even with two or three rings, can outperform square QAM at the same order in a non-linear channel: APSK sacrifices some minimum distance in the linear case to gain much more distance after non-linear compression.

Adaptive Coding and Modulation (ACM) in Real Systems

Static MODCOD selection is the past. Every modern DVB-S2X system uses ACM, where the gateway continuously receives SNR feedback from each remote terminal and adjusts that terminal's forward-link MODCOD on a superframe basis (roughly every 200 to 300 ms in typical deployments).

A realistic ACM policy for a Ka-band consumer broadband forward link might look like this:

• Clear sky, high margin: 32APSK 4/5, spectral efficiency ~3.9 bit/s/Hz
• Light haze or slight rain: 16APSK 3/4, ~3.0 bit/s/Hz
• Moderate rain: 8PSK 3/4, ~2.2 bit/s/Hz
• Heavy rain: QPSK 2/3, ~1.3 bit/s/Hz
• Extreme fade: QPSK 1/4 or DVB-S2X VL-SNR frames, ~0.4 bit/s/Hz

The policy has to include hysteresis so you do not ping-pong between MODCODs, and it has to account for the round-trip latency of the SNR feedback (~500 ms for GEO). For LEO constellations with round-trip latencies under 50 ms, ACM can be much more responsive.

ACM interacts with capacity planning in ways that catch newer engineers off guard. A transponder that closes at 32APSK 5/6 in clear sky delivers roughly twice the bit rate it delivers at 16APSK 2/3 in rain. If 20 percent of your service area is under a rain cell for 5 percent of the year, the aggregate capacity you can sell is not the peak; it is a weighted average that has to satisfy service-level agreements at outage percentiles like 99.5 or 99.9. Operators run stochastic rain-fade simulations against ITU-R P.618 propagation models to size their networks, and the choice of highest and lowest supported MODCODs is a business decision as much as an engineering one.

VCM, CCM, and the Multi-Beam Reality

DVB-S2 defines three carrier modes: constant coding and modulation (CCM), variable coding and modulation (VCM), and adaptive coding and modulation (ACM). CCM is used mostly for broadcast where one MODCOD is chosen to close for the worst-case receiver in the coverage area. VCM assigns different MODCODs to different services on the same carrier without terminal feedback, useful when you know some services (e.g., premium subscribers with larger dishes) can handle higher MODCODs.

ACM is the norm for two-way broadband. On modern high-throughput satellites (HTS) with dozens of spot beams, each beam typically runs its own ACM loop with its own MODCOD distribution, since rain fade is spatially localized. The Ka-band spot beams on a satellite over Europe might have three or four beams degraded at any given moment while the rest are running at peak MODCOD.

Multi-beam operation also introduces a modulation-adjacent decision: precoding. When beams reuse frequency, they interfere with each other. Multi-user MIMO precoding, applied at the gateway, can cancel much of that interference before it hits the air, but it requires the modulator to jointly design symbol streams across beams. DVB-S2X added superframe structures specifically to support beam-hopping and precoding, and this is one of the active research fronts in satcom.

Testing, Verifying, and Debugging Modulation in the Field

A modulation scheme that looks great on a spec sheet can still fail in operations, and the debugging discipline for real links is worth internalizing.

The first-line diagnostic is the modulation error ratio (MER), sometimes called receive constellation quality. MER measures the RMS distance of received symbols from their ideal constellation points, expressed in dB. For DVB-S2 at 16APSK 3/4, you want MER above 12 dB to have any margin over threshold. For 32APSK 4/5, above 14 dB. MER combines every impairment: thermal noise, phase noise, amplifier non-linearity, adjacent-channel interference, and pointing error.

When MER is too low, you have to decompose the causes. Spectrum-analyzer traces of the transmit signal show spectral regrowth, which points to amplifier non-linearity. Constellation-plot inspection shows whether the impairment is rotational (phase noise) or radial (amplitude compression). Bit error rate curves versus Es/N0 tell you whether the receiver is behaving as expected or has a bug in its LDPC decoder.

The waveform side of things also has to be verified in ground testing, and much of that work looks a lot like spacecraft software testing: repeatable test vectors, hardware-in-the-loop simulators (usually a vector signal generator plus a channel emulator plus the flight modem), and pass/fail criteria written against the standard's threshold table. When we cover this in our labs, students trace an entire MODCOD from constellation generation through pulse shaping, upconversion, channel simulation, downconversion, and demodulation, watching where the errors creep in. That is also the kind of hands-on approach our live sessions are designed for: watching an engineer actually iterate on a modem configuration is far more useful than reading a spec table.

Beyond DVB-S2X: What Is Coming Next

DVB-S2X will remain dominant through the late 2020s, but three trends are worth watching.

First, DVB-RCS2 on the return link continues to evolve, with MF-TDMA using QPSK, 8PSK, and 16QAM (yes, QAM on the return, because the ground-terminal SSPA is small and can be linearized more easily than a satellite TWTA). Return links are inherently smaller-carrier, and the modulation choices there are driven more by cost and burst-mode acquisition than by spectral efficiency.

Second, 5G NTN (non-terrestrial networks) as standardized in 3GPP Release 17, 18, and 19 brings OFDM waveforms into satellite links. OFDM has famously bad PAPR, so 5G NTN is either used on regenerative payloads with linear amplifiers or with amplifier pre-distortion and tone-reservation techniques. This is a real shift, because for the first time a mainstream terrestrial waveform is going up to space at scale, and the modulation choices (QPSK, 16QAM, 64QAM, 256QAM per 5G's MCS tables) are being validated in orbit.

Third, optical inter-satellite links operate at completely different modulation orders: on-off keying (OOK), differential phase-shift keying (DPSK), and increasingly coherent QPSK and 16QAM at line rates of 2.5, 10, 25, and 100 Gbps. These do not compete with RF modulation choices; they complement them, since optical links are used between satellites while RF is still used to the ground.

For an engineer entering the field in 2026, the honest advice is: know DVB-S2X cold, understand why APSK dominates over QAM in bent-pipe RF, and start learning 5G NTN and optical link modulation because that is where the next decade of jobs will be.

Building a Career in Satellite Modulation and Link Design

Modulation is one of those specialties where deep understanding transfers well across employers. The math and the standards are the same whether you work at a GEO operator, a LEO constellation, a defense integrator, or an ground-segment vendor. What differs is which part of the MODCOD table you obsess over.

If you are aiming at commercial broadband, get fluent in DVB-S2X ACM, HTS beam economics, and rain-fade statistics. If you are aiming at defense or space-agency work, dig into MIL-STD waveforms, spread-spectrum techniques (which are effectively an additional layer on top of a base modulation), and low-probability-of-intercept design. If you are aiming at the 5G NTN space, learn OFDM PAPR management, timing advance for LEO, and 3GPP release procedures.

Refonte Learning built its satcom track around exactly this map. Students work through link budgets, MODCOD selection, ACM tuning, and hands-on modem configuration, and they finish with an internship on a real project instead of a certificate that sits in a drawer. The full curriculum for the satellite communications engineering program covers modulation as one of about a dozen depth areas, alongside RF link budgets, ground station operations, VSAT networking, and 5G NTN, so you leave with the whole picture rather than a slice of it.

Whatever path you pick, the practical habit that separates good link designers from great ones is the same: always know the current MODCOD, always know the required Es/N0, always know the actual Es/N0, and always know why the difference is what it is. Modulation is not a checkbox on a spec sheet. It is the daily language of the link.

About Refonte Learning

Refonte Learning is an EdTech platform operated by Refonte Infini Infiniment Grand, a French SAS registered under SIREN 949 841 605, with an operational office at 1 Poulton Close, Dover, Kent CT17 0HL, United Kingdom. We run practitioner-led programs in satellite communications, AI, data engineering, cloud, DevOps, and software engineering, each combining live instruction, hands-on labs, and a real internship. If satcom modulation is your entry point, come build the rest of the link with us.