METI, physics, encoding for universality, and new algorithm ideas that close the loop between transmitter design and receiver sensitivity.
Most of this series is about receiving: how MitraSETI ingests spectrograms, removes RFI, de-Dopplers, clusters candidates, and scores interestingness. This chapter turns the telescope around — conceptually — and asks how a civilization (including ours) would transmit a signal meant to be found, decoded, and understood across light-years. That question is not academic fluff. The design of intentional transmissions constrains what "good" technosignatures look like; a receiver that only hunts for one narrow archetype may miss signals optimized under different physics or cultural assumptions. We also survey efficiency: joules per bit, beamforming gain, bandwidth tradeoffs, and coding — then outline new transmission-side ideas that could inspire both future METI experiments and complementary detection modes in MitraSETI.
MitraSETI is a receiver. It detects and analyzes incoming radio energy, turning time–frequency data into candidate events and ranked hypotheses about artificial narrowband structure. The rest of the tutorial series stays on that side of the link budget.
The flip side still matters for four reasons.
They are scientifically adjacent but ethically and institutionally distinct. Many radio astronomers comfortably work on SETI surveys while opposing unilateral powerful METI without broad consultation.
Stephen Hawking and others have publicly argued that announcing our position and capability could carry existential risk if advanced civilizations are not universally benign — an argument from incomplete information about the distribution of motives and technologies in the cosmos. Others counter that Earth has already leaked powerful radar and television signatures, and that coherent METI is a small increment on that background; or that contact could yield enormous benefits.
Institutional framing: The International Academy of Astronautics (IAA) has developed SETI detection protocols (post-detection procedures, verification, consultation). Separate statements and draft protocols have addressed transmission from Earth: the emphasis is typically on international scientific consultation, transparency, and risk-aware deliberation rather than a single national or private actor acting alone. Positions differ among the SETI Institute, METI International, and national agencies; the through-line for responsible practice is governance and consent, not merely technical feasibility. Hawking-style caution and METI advocacy are both compatible with better science: even if humanity never transmits, designing hypothetical beacons clarifies what honest technosignature searches must remain sensitive to.
This chapter treats how transmission works and how algorithms could improve it. Whether any given transmission should occur is a policy and ethics question layered on top of the physics.
Humanity's intentional interstellar messages are few, famous, and instructive: each encodes assumptions about universality (math, chemistry, bitmaps) and about what a receiver might try first (primes, simple grids).
On 16 November 1974, the Arecibo Observatory sent a 1,679-bit binary message toward the globular cluster M13 (Messier 13).
Why 1,679 bits? 1,679 factors uniquely as 23 × 73 (both prime). The hope: a recipient would try rearranging the bit stream into low-dimensional rectangles; only one non-trivial factorization into two primes suggests a 23 × 73 raster — strong evidence of intentionality before any semantic decoding.
The 1,679-bit Arecibo message encoded the numbers 1–10, atomic numbers of life-essential elements (H, C, N, O, P), a schematic of DNA, a crude human figure, a solar system diagram, and the Arecibo dish itself — all in a single 23 × 73 pixel bitmap. It was a three-minute ceremonial broadcast toward a star cluster 25,000 light-years away; the signal won't arrive until roughly calendar year ~26,974 CE.
Radio parameters (commonly cited):
| Quantity | Representative value |
|---|---|
| Frequency | 2380 MHz (2.38 GHz) |
| Duration | ~3 minutes |
| Modulation | Frequency shift (order 10 Hz) between two states to encode binary |
| Power (engineering shorthand) | Often quoted around ~1 MW class effective radiated power; exact published numbers vary slightly by definition (ERP vs EIRP — see §3) |
Distance and arrival: M13 is roughly 25,000 light-years away. Photons travel at c, so the pulse transit time is also ~25,000 years; a 1974 transmission would not wash over the cluster until roughly calendar year ~26,974 CE. The targeting was partly symbolic (a dense stellar environment) rather than an expectation of a near-term reply.
Cosmic Call used Ukraine's Evpatoria RT-70 (70 m) antenna to send structured messages toward nearby Sun-like stars. Cosmic Call 1 (1999) targeted systems such as 16 Cygni A, 15 Sagittae, HD 178428, and Gliese 777. Cosmic Call 2 (2003) added further targets including 47 Ursae Majoris. Related Evpatoria campaigns (e.g. Teen Age Message, 2001–2003 era) used the same facility class with different payloads and target sets.
Representative RF parameters (1999 leg, from project summaries): carrier near 5.01 GHz (often quoted as 5010.024 MHz), FSK with frequency deviations of order ±24 kHz, ~150 kW continuous-wave transmit power into the optics, left-hand circular polarization, and minutes-scale dwell per star — decades round-trip light time for those ~50–70 ly targets.
Unlike the one-shot Arecibo ceremony, Cosmic Call was explicitly campaign-shaped: multiple targets, documented page layouts mixing images and binary primer material, and public-facing participation (names, drawings). Engineering-wise, such campaigns highlight serialization problems — how you packetize a message for minutes-long contacts, repeat headers for robustness, and publish human-readable specifications so Earth-side reproducibility survives archival decay.
For MitraSETI practitioners, the takeaway is procedural: any transmission algorithm should ship with a machine-readable description (symbol rate, FEC, modulation class) alongside the human outreach narrative.
"A Message from Earth" (AMFE), tied to the social network Bebo, collected 501 user images and text snippets and transmitted them from Evpatoria's RT-70 toward Gliese 581 in October 2008 — when Gliese 581 c was still prominent in public discourse as a candidate habitable world. At ~20 light-years, the leading edge of the signal arrives Earth-calendar ~2028–2030. Public accounts describe a multi-hour broadcast and high effective power via the large dish. Culturally, AMFE is crowdsourced METI; information-theoretically, it is not optimized for minimal energy per bit or maximal a priori decodability. It remains a useful case study in METI as media versus METI as controlled link experiment.
Jiang et al. (2022) proposed an updated binary bitmap message sometimes called "A Beacon in the Galaxy," intended for discussion with facilities such as FAST (Five-hundred-meter Aperture Spherical Radio Telescope). Compared to the 1974 design, modern proposals typically add:
Targets discussed in such proposals often emphasize directions toward the Galactic Center or selected star fields with favorable column density and science interest — still a normative choice, not a unique optimum.
For an isotropic radiator, power P spreads over a sphere of area 4πr². The flux density (power per unit area) at distance r is:
A directive antenna concentrates power into a solid angle Ω; observers in the main beam see an effective isotropic radiated power (EIRP) much larger than the feed power Ptx:
Effective Isotropic Radiated Power answers: "What isotropic source would produce this flux in the main lobe?"
where G is antenna gain (dimensionless, relative to an isotropic radiator). Historical dishes like Arecibo achieved ~70 dBi (G ~ 10⁷) for planetary radar work.
Worked intuition at 10 light-years: One light-year ≈ 9.461 × 1015 m, so 10 ly ≈ 9.46 × 1016 m. Then:
That distinction matters when you read popular articles: feed power, ERP, and EIRP are different rungs on the same ladder.
Radio astronomers often reason with a radiometer equation flavor: for a resolved narrowband signal of flux density S (W·m⁻²·Hz⁻¹) or equivalently power S·Aeff captured by effective area Aeff, the signal-to-noise ratio (SNR) after integrating for time t in bandwidth B scales roughly as:
(assuming Gaussian noise statistics, matched filtering, and coherent integration over t — details vary with modulation and systematics). Doubling integration time buys √2 in SNR for broadband noise; halving bandwidth B also improves SNR if the signal stays fully inside the channel. Interstellar METI is therefore a three-way negotiation among EIRP (via S), collecting area (Aeff), and observing strategy (t, B, Tsys). MitraSETI's de-Doppler machinery effectively recovers coherence against drift so that t can be long without smearing the tone into thermal noise.
Historical dishes like Arecibo achieved enormous gain — often quoted ~70 dBi (i.e. G ~ 10⁷) for planetary radar–class work, though exact values depend on frequency and illumination. If Ptx = 1 kW and G = 10⁷, then:
With higher feed power or slightly different assumptions, EIRP estimates in the TW range appear in the literature for narrow-beam planetary radar. The lesson for MitraSETI: detection sensitivity must be discussed with an assumed EIRP, distance, bandwidth, and integration time together.
A useful order-of-magnitude relation ties collecting area, system temperature, and integration to detectability. Let Aeff be the effective aperture of the receiving antenna, η an efficiency factor, and S the flux density at the telescope. The received power is Prec ≈ η Aeff S. For broadband noise of equivalent temperature Tsys in bandwidth B, noise power is Pnoise ≈ kB Tsys B. The instantaneous SNR is therefore SNR₀ ≈ Prec / Pnoise. For a coherent tone integrated τ seconds, noise variance in a matched filter scales roughly as 1/τ (power SNR grows linearly with τ for a stable carrier in idealized Gaussian noise). Combining:
(omit constants of order unity; real systems add atmospheric loss, spillover, quantization, RFI, and scintillation). This single proportionality explains the cultural obsession with bigger dishes, cryogenic front ends (lower Tsys), narrower B, and longer dwells — and why a 10⁻²² W/m² class signal is not absurd if Aeff is 10⁴ m² and τ is hours. MitraSETI's algorithms are trying to recover coherence (in frequency and drift) so that effective τ can be stretched without smearing the line into thermal noise.
For additive noise with approximately flat spectral density, the noise power in bandwidth B scales like k Tsys B (receiver noise temperature Tsys). Holding EIRP fixed:
For a circular aperture of diameter D observing at wavelength λ, the Airy-scale beam is θ ≈ 1.22 λ/D radians. Higher frequency (smaller λ) for fixed D yields a tighter main beam — more EIRP on-axis for the same feed power — at the cost of pointing and acquisition difficulty. At interstellar distances, even arcsecond beams cover AU-scale footprints, but stellar proper motion and reflex wobble still matter over years of round-trip light time. Transmitters therefore care about ephemeris updates; receivers care about drift and acceleration — the same physics MitraSETI already parameterizes.
Modulation maps bits (or symbols) onto physical observables: amplitude, frequency, phase, or time of arrival.
An unmodulated single tone is the simplest artifact to find with narrowband SETI: integrate long enough, search over frequency and drift. Many surveys are CW-centric for good reason — high matched-filter gain against broadband noise.
Matched filtering (intuition): Under additive white Gaussian noise, the optimal linear detector for a known waveform correlates the data with a template of that waveform. For a pure sinusoid of unknown phase, the incoherent combination of I/Q correlators recovers power SNR improvement proportional to integration time. That is why Taylor trees and friends exist: the template is not perfectly known because Doppler smears frequency; the search enumerates plausible drift tracks to re-cohere energy before applying the narrowband matched filter mentally implemented as FFT stacks and path integration.
OOK sends power on for "1" and (nearly) off for "0." Pros: trivial demodulation; cons: the off intervals carry no energy — average power is wasted relative to constant-envelope schemes if duty cycle < 1.
FSK uses two (or more) frequencies for symbols. Envelope can stay on; spectral occupancy widens.
PSK encodes bits in phase discontinuities of a carrier. BPSK uses two phases (0° / 180°); QPSK uses four phases (2 bits/symbol). For many channels, PSK families approach good power efficiency at a given bit error rate compared to simple OOK.
PPM places energy bursts in one of several time slots per symbol period. It can be energy-efficient at low duty cycle because the peak power is high but average may be moderate — useful when peak EIRP buys detectability but mean power is constrained.
David Messerschmitt analyzed energy-efficient interstellar signaling under the assumption that bandwidth is cheap compared to energy (photons delivered at the receiver). Location coding maps symbols to time–frequency coordinates of short bursts — approaching minimum energy per bit in idealized models by avoiding long continuous carrier power when not needed. For SETI detection, this suggests burst searches in spectrograms, not only constant tones.
CW and slow OOK align with narrowband hit finding + drift tracking. FSK/PSK require different matched filters (multi-hypothesis frequency tracks, phase tracking). Location codes push toward sparse burst clustering in time–frequency and higher-dimensional feature extractors — an algorithmic frontier, not a parameter tweak.
A METI designer assumes no shared natural language. The message must bootstrap understanding from physics and logic.
Two symbols are the default digital substrate — easy to modulate, easy to threshold.
Listing primes (2, 3, 5, 7, 11, …) is a standard attention-getter: low Kolmogorov complexity for the concept, unlikely as a narrow astrophysical line process, and culturally neutral.
Lincos (Hans Freudenthal, 1960) is a constructed cosmic language built from logic, arithmetic, and staged dialog — designed to teach semantics by examples. It influenced later cosmic language thinking even when impractical for short beacons.
2D rasters (Arecibo, modern proposals) trade human interpretability for ambiguity (which axis is time? color depth?). Still, geometry and counting bridge cultures.
Algorithmic or executable messaging sends a program whose output is the payload (sometimes called ACETI-style ideas in the literature). Pros: compressed description of complex objects; Cons: requires a shared machine model (but Turing-complete models are universal in CS). Security note: arbitrary code is dangerous on Earth systems; the cosmic version is hypothetical — but sandboxed emulation is the obvious Earth-side discipline.
A message can embed its own decoder specification in layers: layer 0 is a simple amplitude pattern; layer 1 describes symbol rate; layer 2 specifies decompression. This mirrors human file formats (headers + payload).
Fibonacci, π digits, prime gaps — structured but non-linguistic — can serve as synchronization and "wow" factors. They are not proof of intelligence alone (pulsars were once "LGMs"), but in concert with non-natural modulation they strengthen posterior odds.
For thermal noise at temperature T, there is a famous minimum energy per information bit scale:
The minimum energy per information bit at capacity, idealized for thermal noise:
At T = 300 K, kBT ln 2 ≈ 2.9 × 10⁻²¹ J — tiny. Reality: interstellar links are power-limited and photon-starved; path loss, not Johnson noise in the receiver front end, often dominates the engineering SNR. Still, Shannon sets a long-term ceiling: codes like LDPC and polar codes approach capacity on many channels, translating SNR into reliable bits/s/Hz.
When bandwidth is abundant, spread energy thinly and pulse rarely. Direct-sequence spread spectrum (DSSS) multiplies a narrow data waveform by a wide pseudo-noise (PN) chip sequence at chip rate Rc ≫ symbol rate Rs. The processing gain is roughly Gp ≈ Rc / Rs: the spread signal sits below the noise floor in a naive wideband detector until despread with the correct PN. For METI, the design trick is making the PN derivable from mathematics rather than a secret key — otherwise no unknown civilization can ever find you.
Frequency hopping jumps a narrow carrier among M slots according to a sequence fk = f₀ + Δf · h(k). An autocorrelation search across lag and hop hypothesis is heavier than CW search but parallelizes well — an interesting GPU generalization of the Taylor tree beyond linear drift in a single frequency channel.
Phased arrays steer beams by phase-shifting signals from N elements. Ideal coherent combination yields field amplitude ∝ N and power ∝ N² in the main lobe for equal amplitudes:
(scaling simplified; grating lobes and calibration matter). Example: N = 100 elements each at 10 W coherent effective radiation can produce dramatic peak EIRP compared to a single 10 W isotropic source — why SETI arrays and radar communities invest in calibration and digital beamforming.
Transmit when:
This is operations research on top of RF engineering.
Kolmogorov complexity measures intrinsic description length; arithmetic coding approaches entropy for known symbol statistics. But maximally compressed streams look like random noise unless the receiver already knows the codebook and model. Practical METI therefore often uses:
The following items are speculative engineering meant to seed real implementations — some extend MitraSETI's receiver logic; others are transmitter-side only.
Problem: The transmitter does not know the receiver's exact velocity and acceleration relative to the line of sight; binary stars, exoplanets, and stellar reflex motion add curvature to drift.
Status quo: Transmit near-constant frequency; the receiver runs Taylor trees or chirp search.
Proposal: Embed a known chirp law f(t) = f₀ + ḟ t + ½ f̈ t² + ⋯ as part of the protocol. The signal is self-calibrating: a matched filter for that polynomial in frequency yields higher SNR than a generic wide drift search if the pattern is correct.
Concrete example: Sweep 100 Hz over 10 minutes (ḟ ≈ 100 Hz / 600 s ≈ 0.167 Hz/s) with quadratic residuals below 0.1 Hz over the pass — publish the coefficients in metadata (for humans) and repeat them in binary prelude (for ETI).
Detection gain (order of magnitude): A matched filter to a known chirp integrates coherent energy along a one-dimensional track in (f, t); a generic drift search must tile many (f₀, ḟ, f̈) triplets. If the chirp hypothesis space without prior has N viable templates and the true waveform is one of them, coarse search pays a ~√N penalty in effective SNR relative to a single matched filter (intuition: energy is split across trials / multiple-testing correction). Publishing the chirp law collapses N → 1 for civilizations that understand the protocol — and gives Earth-side MitraSETI a labeled training distribution for chirp-aware classifiers.
MitraSETI tie-in: The repository already prototypes acceleration-aware search in scripts/chirp_search.py, which extends linear de-Doppler to second-order drift and scores improvements with vs without chirp correction. A transmission standard could deliberately emit chirps that this machinery excels at finding — closing the loop between METI waveform and SETI software.
Problem: Narrowband beacons are fragile: a localized RFI event or notch filtering can erase a single tone.
Proposal: Direct-sequence or frequency-hopped waveforms keyed by a public mathematical object:
Receiver discovery path: First detect comb structure in wideband autocorrelation; then despread with the hypothesis class of prime-indexed hop patterns.
Concrete pattern example (hydrogen × primes "comb"): Define hop carriers fk = fHI · pk / p₁ for small primes pk ∈ {2, 3, 5, 7, 11, …} and fHI ≈ 1420.405 MHz. Each hop dwell lasts Th seconds; the entire cycle repeats with period L · Th for L hops. A receiver without the full spreading key might still notice a comb in wideband spectra whose ratios approximate rational combinations of primes — a geometric "hello" orthogonal to single-line pulsars but not generic Gaussian noise.
Micro-timing variant: Prime-weighted slotting (Δtk ∝ 1/pk) yields sparse periodicities in spectrogram autocorrelation before full despreading; a second stage tests small public libraries (M-sequences from published primitive polynomials, Walsh blocks). Hypothesis explosion naturally points to GPU correlation banks and HDBSCAN-style peak linking (tutorial 05).
Novelty: Consumer SETI stacks rarely implement CDMA-style interstellar search; this is genuinely under-explored relative to CW surveys.
Concrete detection sketch: Partition wideband data into chunks; build magnitude-squared spectrograms and coherently sum along low-parameter hop templates h(k) (e.g. h(k) = p(k mod Nprimes) mod M with published Nprimes). Peaks that survive for algebraic templates but not random permutations are candidates. A MitraSETI extension could ship template banks analogous to GW matched-filter banks, but in time–frequency rather than strain.
Quantum error-correcting codes inspire classical parity-check constructions; for METI, the actionable near-term tools are capacity-approaching classical codes:
Recommendation for a MitraSETI "transmitter module": implement polar-coded frames with explicit frozen-bit pattern published in Earth documentation; demodulation uses SC or SC-list decoding — test at SNR thresholds where uncoded BPSK fails. This is not "quantum METI"; it is modern classical coding framed for deep-space-class links.
Simulation recipe: Fix code length N = 2n (e.g. n = 10 → N = 1024), rate R = K/N, BPSK on AWGN. Sweep Eb/N₀ and compare frame error rate for uncoded, (255,239) Reed–Solomon outer + convolutional inner (historical CCSDS flavor), LDPC, and polar with CRC-aided list decoding. Plot required Eb/N₀ at FER = 10⁻³. You will see why deep-space missions moved from Voyager-era codes toward capacity-approaching families — and why a METI designer might still choose simple repetition for the beacon and polar only for the payload after lock.
Concept: Simultaneous radio and optical (or IR laser) pulses with known time–frequency offsets. A coincidence requirement slashes false alarms from terrestrial RFI (radio-only) or satellite glints (optical-only).
MitraSETI + AstroLens analogy: Unified multi-wavelength reasoning on reception (e.g. correlated sky maps) has a transmission dual: coordinated emitters with public ephemerides.
Implementation sketch: A radio array transmits a BPSK beacon; a laser facility fires ns-class pulses locked via GPS-disciplined clocks; both encoders share a TAI-based timestamp block in the header.
False-alarm arithmetic (toy model): Suppose a radio search has false-alarm probability Pfa,r per trial and an optical search Pfa,o, with independence conditional on noise. A coincidence requirement within Δt and Δθ on the sky yields Pfa ≈ Pfa,r × Pfa,o × Nc, where Nc counts accidental alignments in time–angle cells. Even if Pfa,r and Pfa,o are only moderately small — say 10⁻³ each — joint 10⁻⁶-scale accidentals can be achieved without heroic single-channel thresholds. That is the scientific motivation for multi-messenger beacons: not mysticism, but correlated structure across independent noise processes.
Layer 0 — Acquisition: High EIRP CW or slow OOK for discovery with existing SETI pipelines.
Layer 1 — Spec sheet: Low-rate BPSK describing symbol timing, forward error correction, and a "build this aperture" diagram — still bitmap-like.
Layer 2 — High-rate payload: Wideband PSK with polar/LDPC coding, decodable only after sufficient G/T (gain over system temperature).
Interpretation: The message is not one file — it is an upgrade path for the receiver civilization. Ethically, this is loaded: it suggests technological influence. As pure information theory, it mirrors progressive JPEG or HTTP range requests: bootstrap then stream.
Problem: Matched filtering a single pseudorandom burst against itself produces range (time-delay) sidelobes — false peaks that look like echoes of the beacon. In interstellar searches, interstellar plasma and multipath-like scattering already complicate delay structure; high sidelobes waste follow-up time.
Proposal: Transmit pairs of Golay complementary sequences A and B (same energy, designed so autocorrelation sidelobes cancel when summed). Alternate A and B bursts on the same carrier (or hop them as in §7b). A receiver correlates separately, then adds the magnitude-corrected correlation outputs; ideal models yield very low range sidelobes for the composite.
Why it is METI-relevant: Golay pairs are public mathematical objects — no shared secret — and the decoder is a linear operation (two FIR correlations + sum), plausible for any civilization that discovers discrete signal processing. Data rate is modest: you pay ~2× time for the pair relative to one PN burst.
MitraSETI path: Extend burst clustering to look for alternating templates (+1, −1) weighting on odd/even burst indices; score peak sharpness in delay domain vs single-sequence hypotheses. This is not implemented in typical CW pipelines and pairs naturally with location codes (§4).
A symmetric architecture to the receiving pipeline might contain:
Software stack (prototype-friendly): Python + NumPy for waveform synthesis; GNU Radio for SDR I/Q streams; hardware restricted to legal bands and power (see §10).
This module would not replace national observatories; it would standardize test waveforms that MitraSETI (and partners) can simulate, transmit at mW scale, and re-acquire — turning METI research into reproducible RFML experiments.
You do not need a sky-filling dish to study algorithms.
Workflow:
Hardware shakedown: RTL-SDR, HackRF, USRP, or similar SDRs in ISM or licensed bands at low power validate clocking, filtering, and nonlinearities missing from pure simulation.
Chapter 11 expands hands-on testing; treat this section as the research motivation for those labs.
International law does not provide a crisp prohibition of METI analogous to nuclear test bans; practice is guided by norms, institutional policies, and spectrum regulation. The absence of a binding treaty shifts responsibility to institutions and professional ethics.
Transmission research does not replace passive SETI; it informs it. Efficient waveforms may look less like naive CW and more like structured chirps, sparse bursts, or spread patterns — each implying detector upgrades. MitraSETI sits in a useful position: it can simulate, search, and score those hypotheses without immediately pointing a megawatt dish at a neighbor star. If humanity ever does transmit with serious intent, let the waveform be as thoughtful as the ethics — and let our receivers be ready for both the beacon we hope to find and the standards we choose to broadcast.
scripts/chirp_search.py for quadratic drift / chirp-rate search aligned with §7a.This document is instructional research, not legal advice. Consult qualified counsel and regulators before any on-air transmission.