PRATUSH: Lunar 21-cm Radiometer

• PRATUSH is a proposed lunar orbiter radiometer designed to detect the global redshifted 21‐cm signal from the Cosmic Dawn and Epoch of Reionization.
• It employs a single-payload instrument with a frequency-independent monocone antenna and in-situ impedance measurements to achieve milliKelvin-level spectral precision.
• The mission uses lunar farside operations in a radio-quiet prime cone and Maximally Smooth foreground modeling to isolate weak cosmological signals from strong terrestrial interference.

PRATUSH—Probing ReionizATion of the Universe using Signal from Hydrogen—is a proposed, solo-payload lunar-orbiter radiometer from the Raman Research Institute designed to make a precision measurement of the absolute, sky-averaged radio spectrum in the 40–200 MHz band, optimized for 55–110 MHz, in order to detect the global redshifted 21-cm signal from the Cosmic Dawn and Epoch of Reionization. Its baseline concept couples a purpose-built, spectrally smooth instrument to the radio-quiet environment above the lunar farside, where science observations are made when the spacecraft is behind the Moon relative to both Earth and Sun, with downlink on the near-side. PRATUSH was proposed to the Indian Space Research Organization in 2018 and is in the pre-project studies phase (Rao et al., 8 Jul 2025).

1. Scientific aim and cosmological observable

The scientific goal of PRATUSH is to detect the global, or monopole, 21-cm signal: the sky-averaged differential brightness temperature $\delta T_b(z)$. In the baseline formulation, this observable directly probes the onset of first stars and galaxies, X-ray heating of the intergalactic medium, and the subsequent reionization history. The signal brightness relative to the background radiation is written as

$$\delta T_b = 27\, x_{\mathrm{HI}} (1 + \delta_b)\left(\frac{\Omega_b h^2}{0.023}\right) \left[\frac{0.15}{\Omega_m h^2}\times \frac{1+z}{10}\right]^{1/2} \times \frac{T_s - T_R}{T_s} \times \left[\frac{\partial_r v_r}{(1+z)H(z)}\right]\ \mathrm{mK},$$

where $x_{\mathrm{HI}}$ is the neutral fraction, $T_s$ the spin temperature, $T_R$ the radiation temperature, $\delta_b$ the baryon overdensity, and the last factor accounts for line-of-sight velocity gradients. The spectral turning points in $\delta T_b(\nu)$ encode when Ly-$\alpha$ coupling, X-ray heating, and ionization take hold, thus constraining star formation efficiency, X-ray production, and reionization timing (Rao et al., 8 Jul 2025).

PRATUSH adopts the standard frequency-redshift mapping

$z(\nu) = \frac{1420.40575\ \mathrm{MHz}}{\nu} - 1.$

Within the baseline band, $55\ \mathrm{MHz} \rightarrow z \approx 24.8$, $$\delta T_b = 27\, x_{\mathrm{HI}} (1 + \delta_b)\left(\frac{\Omega_b h^2}{0.023}\right) \left[\frac{0.15}{\Omega_m h^2}\times \frac{1+z}{10}\right]^{1/2} \times \frac{T_s - T_R}{T_s} \times \left[\frac{\partial_r v_r}{(1+z)H(z)}\right]\ \mathrm{mK},$$0, and $$\delta T_b = 27\, x_{\mathrm{HI}} (1 + \delta_b)\left(\frac{\Omega_b h^2}{0.023}\right) \left[\frac{0.15}{\Omega_m h^2}\times \frac{1+z}{10}\right]^{1/2} \times \frac{T_s - T_R}{T_s} \times \left[\frac{\partial_r v_r}{(1+z)H(z)}\right]\ \mathrm{mK},$$1. The 55–110 MHz interval was selected because it spans key Cosmic Dawn turning points with strong spectral leverage, contains the terrestrial FM band from 88–108 MHz, and targets earlier Cosmic Dawn epochs that remain comparatively less constrained by ground experiments. Standard models predict features with amplitudes of approximately $$\delta T_b = 27\, x_{\mathrm{HI}} (1 + \delta_b)\left(\frac{\Omega_b h^2}{0.023}\right) \left[\frac{0.15}{\Omega_m h^2}\times \frac{1+z}{10}\right]^{1/2} \times \frac{T_s - T_R}{T_s} \times \left[\frac{\partial_r v_r}{(1+z)H(z)}\right]\ \mathrm{mK},$$2–$$\delta T_b = 27\, x_{\mathrm{HI}} (1 + \delta_b)\left(\frac{\Omega_b h^2}{0.023}\right) \left[\frac{0.15}{\Omega_m h^2}\times \frac{1+z}{10}\right]^{1/2} \times \frac{T_s - T_R}{T_s} \times \left[\frac{\partial_r v_r}{(1+z)H(z)}\right]\ \mathrm{mK},$$3. For sensitivity tests, PRATUSH uses a fiducial Gaussian absorption with amplitude $$\delta T_b = 27\, x_{\mathrm{HI}} (1 + \delta_b)\left(\frac{\Omega_b h^2}{0.023}\right) \left[\frac{0.15}{\Omega_m h^2}\times \frac{1+z}{10}\right]^{1/2} \times \frac{T_s - T_R}{T_s} \times \left[\frac{\partial_r v_r}{(1+z)H(z)}\right]\ \mathrm{mK},$$4 centered at $$\delta T_b = 27\, x_{\mathrm{HI}} (1 + \delta_b)\left(\frac{\Omega_b h^2}{0.023}\right) \left[\frac{0.15}{\Omega_m h^2}\times \frac{1+z}{10}\right]^{1/2} \times \frac{T_s - T_R}{T_s} \times \left[\frac{\partial_r v_r}{(1+z)H(z)}\right]\ \mathrm{mK},$$5 with $$\delta T_b = 27\, x_{\mathrm{HI}} (1 + \delta_b)\left(\frac{\Omega_b h^2}{0.023}\right) \left[\frac{0.15}{\Omega_m h^2}\times \frac{1+z}{10}\right]^{1/2} \times \frac{T_s - T_R}{T_s} \times \left[\frac{\partial_r v_r}{(1+z)H(z)}\right]\ \mathrm{mK},$$6.

The astrophysical interpretation of a detection is central to the concept. The turning points and depths constrain Ly-$$\delta T_b = 27\, x_{\mathrm{HI}} (1 + \delta_b)\left(\frac{\Omega_b h^2}{0.023}\right) \left[\frac{0.15}{\Omega_m h^2}\times \frac{1+z}{10}\right]^{1/2} \times \frac{T_s - T_R}{T_s} \times \left[\frac{\partial_r v_r}{(1+z)H(z)}\right]\ \mathrm{mK},$$7 coupling, identified with the onset of first stars, the X-ray heating efficiency of early sources, and the timing of reionization. With two-year sensitivity and mK-level spectral precision after foreground subtraction, detections or limits are intended to translate to constraints on star-formation efficiency, X-ray luminosity per stellar mass, and reionization history parameters, complementary to power-spectrum experiments.

2. Lunar-farside operations and orbital logic

The mission concept places PRATUSH in lunar orbit and restricts science observations to the radio-quiet region behind the Moon relative to both Earth and Sun, termed the “prime cone.” This geometry is intended to maximize radio quietness by excluding direct sightlines to the dominant contaminating emitters. Data are downlinked to Earth ground stations when the spacecraft is on the near-side, and batteries are charged via solar panels when the spacecraft is sunlit. The baseline orbit strategy is a low-inclination lunar orbit with an average orbital period of approximately 2 hours; the design study notes that specific altitude and inclination remain mission trades, and references frozen orbits as an option. Multiple orbits were studied, with the adopted choice maximizing time in the prime cone with minimal station-keeping (Rao et al., 8 Jul 2025).

A conservative estimate is that 15% of total mission time is “science useful,” after accounting for electronics stabilization, calibration, and operational constraints. Over a 2-year mission, PRATUSH requires approximately 200 hours of integration within the prime cone to achieve sensitivity goals in the 55–110 MHz band. The operational sequence per prime-cone pass is explicit: wake up and thermal stabilization, in-situ vector network analyzer measurement of antenna return loss $$\delta T_b = 27\, x_{\mathrm{HI}} (1 + \delta_b)\left(\frac{\Omega_b h^2}{0.023}\right) \left[\frac{0.15}{\Omega_m h^2}\times \frac{1+z}{10}\right]^{1/2} \times \frac{T_s - T_R}{T_s} \times \left[\frac{\partial_r v_r}{(1+z)H(z)}\right]\ \mathrm{mK},$$8 before and after the observing pass, bandpass-calibrated sky measurements, and then sleep mode. Downlink occurs on the near-side, and charging occurs on sunlit legs.

The lunar farside is chosen because it is shielded from terrestrial radio-frequency interference and has essentially no ionosphere. The mission rationale emphasizes that measurements from RAE-2 showed the Moon attenuates Earth RFI at long wavelengths, and simulations indicate shielding extends into the PRATUSH bands. The orbital platform also avoids ionospheric refraction, ionospheric absorption, and horizon or ground coupling systematics intrinsic to ground experiments. The inclusion of the FM band in the baseline science range is therefore not incidental: the same 88–108 MHz interval that is a critical RFI contaminant on Earth is specifically part of the spectral region PRATUSH seeks to exploit.

A separate laboratory-model paper describes PRATUSH as a two-phase mission, with Phase I in Low-Earth Orbit and Phase II in lunar orbit. In that formulation, the Low-Earth Orbit phase serves as a technology demonstration and as a precursor to lunar deployment, whereas the lunar farside remains the environment identified as best suited to achieving milliKelvin-level residuals (S. et al., 8 Jul 2025).

3. Instrument architecture and design principles

PRATUSH is designed to operate as a solo experiment with a dedicated spacecraft. The explicit reason is the highly sensitive nature of the measurement: a single-payload platform reduces interference from co-manifested payloads and enables tighter control of spacecraft-generated electromagnetic interference. The baseline antenna is a wideband, frequency-independent monocone over a shaped reflector, integrated with a bus of dimension $$\delta T_b = 27\, x_{\mathrm{HI}} (1 + \delta_b)\left(\frac{\Omega_b h^2}{0.023}\right) \left[\frac{0.15}{\Omega_m h^2}\times \frac{1+z}{10}\right]^{1/2} \times \frac{T_s - T_R}{T_s} \times \left[\frac{\partial_r v_r}{(1+z)H(z)}\right]\ \mathrm{mK},$$9, approximately $x_{\mathrm{HI}}$0. The reflector has an optimized log-spiral profile to produce a smooth return loss from 55–110 MHz. The beam is azimuthally symmetric and kept as chromaticity-free as practical; spectral smoothness is validated via simulations and a dedicated pipeline (Rao et al., 8 Jul 2025).

The radiometer front end routes the antenna feed to a “master” RF switch that selects either observation mode, into the analog receiver, or antenna characterization mode, into the vector network analyzer. Receiver calibration is based on double-differenced Dicke switching with a flat-spectrum noise source and a reference attenuator. The 6-state switching cycle uses a phase switch $x_{\mathrm{HI}}$1, a Dicke switch, and noise ON/OFF. The calibrated spectrum per cycle is

$x_{\mathrm{HI}}$2

where $x_{\mathrm{HI}}$3 is obtained from absolute hot/cold load calibration.

Reflection control is a first-order design constraint. The standing-wave spacing is

$x_{\mathrm{HI}}$4

so for $x_{\mathrm{HI}}$5 and velocity factor approximately $x_{\mathrm{HI}}$6, $x_{\mathrm{HI}}$7. PRATUSH therefore keeps cables between cascaded components to centimeter scale, so that residual ripples are effectively DC-like over 55–110 MHz. Thermal control is likewise explicit: the RF electronics and digital receiver are enclosed within thermally regulated, multi-layer EMI-shielded chassis inside the bus, while the antenna and reflector are externally exposed and tracked in flight for thermal and plasma-induced impedance drifts via the VNA. Electromagnetic compatibility requirements are correspondingly stringent, with a required shielding effectiveness $x_{\mathrm{HI}}$8 for digital electronics, compared with $x_{\mathrm{HI}}$9 for the team’s ground-based SARAS.

The flight-baseline digital receiver uses two 12-bit ADCs (EV12AD550B) sampling at 250 MSps and a space-grade Virtex-5QV FPGA (XQR5VFX130) implementing an FX cross-correlation spectrometer with a 2048-point FFT, 4-term Nuttall window, 244 kHz spectral resolution, and 134 ms on-chip integration time. Output products are averaged cross-power spectra with state control and metadata, buffered and reduced by an on-board SBC. The stated power is approximately 35 W, and the operating temperature range is $T_s$0–$T_s$1.

Element	Baseline specification
Measurement range	40–200 MHz goal; 55–110 MHz baseline
Antenna	Frequency-independent monocone over shaped reflector
Digital backend	12-bit ADCs at 250 MSps; 2048-point FFT; 244 kHz
On-chip integration	134 ms
EMI target	$T_s$2 dB shielding effectiveness
Spacecraft mode	Solo, dedicated spacecraft

These architectural choices are organized by a small set of fundamental design principles: stability and smooth spectral response dominate all trades; instrument bandpass smoothness is quantified by Maximally Smooth functions; beam chromaticity and return-loss ripples are minimized; absolute calibration and in-situ impedance characterization are required; reflections are controlled through short cables and good impedance matching; thermal stability is prioritized; and multi-level EMI shielding is treated as a mission-level constraint.

4. Calibration, foreground subtraction, and systematic control

PRATUSH separates absolute calibration from relative bandpass calibration. The absolute calibration factor $T_s$3 is obtained using known hot and cold reference loads. Relative calibration uses the 6-state Dicke switching cycle with a stable in-band noise source and a phase-switched, two-arm correlation receiver, so that the double differencing suppresses additive and multiplicative systematics. This calibration scheme is coupled to explicit in-situ impedance measurement: a one-port vector network analyzer, optimized for 55–110 MHz, measures $T_s$4 across the band, with calibration based on precision open, short, and 50 $T_s$5 terminations together with a reference loop-back path (Rao et al., 8 Jul 2025).

Antenna mismatch enters through

$T_s$6

with the effective antenna temperature

$T_s$7

The use of measured $T_s$8 enables offline correction for impedance mismatch. This is not an ancillary refinement but a requirement of the signal-extraction strategy. The antenna-validation pipeline tests whether beam chromaticity and return loss, after $T_s$9 correction, leave residuals below mK after Maximally Smooth fitting. The numerical result cited for the baseline validation is specific: without $T_R$0 correction, an MS fit to foregrounds leaves approximately $T_R$1 RMS residuals; with $T_R$2 correction, PRATUSH cleanly distinguishes the presence or absence of a Cosmic Dawn signal in simulations.

Foreground modeling is based on GMOSS, an all-sky, physically motivated model that combines synchrotron and other components rather than adopting a single power law for science analysis. The beam-weighted antenna temperature is

$T_R$3

Foreground removal then uses Maximally Smooth functions, constrained so that their higher-order derivatives do not change sign. The logic of the method is that diffuse foregrounds over sufficiently wide bandwidths are spectrally smooth and contain no turning points, whereas the cosmological feature is not Maximally Smooth. In PRATUSH, MS functions therefore serve both as a foreground model and as an instrument-smoothness requirement.

The data-processing chain follows this logic closely. In orbit, state-cycled cross-power spectra are acquired with a 4-term Nuttall window to minimize spectral leakage, together with VNA sweeps. Channel excision is applied around known spacecraft tones placed at FFT bin centers, combined with robust windowing and flagging of outlier channels. The 6-state double-difference calibration is then applied to form $T_R$4, after which the spectrum is corrected for antenna mismatch using measured $T_R$5. Foregrounds are fit with MS functions, residuals are inspected, and Bayesian parameter inference for a 3-parameter Gaussian 21-cm signal is performed with PolyChord. Mock data based on GMOSS convolved with the antenna beam, plus thermal noise from the radiometer equation, with and without an injected 21-cm signal, are used to validate the pipeline.

5. Sensitivity forecasts and signal recovery

The PRATUSH noise budget is sky dominated. In the 55–110 MHz band, the system temperature is dominated by sky synchrotron, and $T_R$6 is computed using the GMOSS sky model convolved with the antenna beam. Thermal sensitivity is estimated with the radiometer equation

$T_R$7

where $T_R$8 is the channel bandwidth and $T_R$9 the total integration time. The baseline parameters are a channel width of approximately $\delta_b$0 and a total prime-cone integration of approximately $\delta_b$1 over two years. Under these assumptions, the thermal noise RMS is at the milliKelvin level across the 55–110 MHz band (Rao et al., 8 Jul 2025).

Signal-detection tests are carried out with mock datasets composed of GMOSS foregrounds convolved with the beam, thermal noise, and optionally a Gaussian 21-cm absorption feature. The model used in simulations and inference is

$\delta_b$2

with $\delta_b$3. For the fiducial case, $\delta_b$4, $\delta_b$5, and $\delta_b$6. The reported result is that the residual RMS increases significantly when the signal is present, indicating detection at mK precision, and PolyChord recovers unbiased estimates of the amplitude, center, and width.

PRATUSH also specifies a stability criterion on calibrated termination measurements: time-averaged, bandpass-calibrated spectra measured against standard terminations must be modeled by a Maximally Smooth function with nearly Gaussian residuals at $\delta_b$7 levels. The significance of this requirement is methodological rather than merely instrumental. If the receiver is thermal-noise dominated and free of confusing spectral features under standard terminations, then the subsequent separation of a non-smooth cosmological feature from smooth foreground structure becomes technically credible.

A plausible implication is that the main limiting factor is not raw radiometric sensitivity but the control of low-level spectral structure. This interpretation is consistent with the repeated emphasis on beam chromaticity, reflection suppression, in-situ impedance tracking, and MS-based validation throughout the baseline design.

6. Laboratory implementation and Earth-orbit precursor studies

A laboratory-model paper presents the digital correlation spectrometer for PRATUSH as an integral subsystem enabling phase switching, digitization, and generation of the sky spectrum. In that implementation, the laboratory model uses 10-bit EV10AQ190 ADCs and a Virtex-6 FPGA, while a Raspberry Pi 4 Model B-based single-board computer acts as the master controller, real-time processor, and data recorder. The firmware implements an FX correlator with a minimum 4-term Nuttall window, $\delta_b$8 spectral channels, effective spectral resolution $\delta_b$9, and on-chip averaging of 2,048 FFT frames for approximately 134 ms integration per output spectrum. The six-state calibration cycle produces 96 spectra per cycle, 16 per state, with a typical cadence of approximately 48 seconds per full cycle (S. et al., 8 Jul 2025).

The laboratory-model enclosure is sized to $\delta T_b(\nu)$0, with a digital system mass of approximately 15 kg and power of approximately 75 W. The paper reports that volume, weight, and power were reduced by approximately 40% relative to a laptop-based SARAS3 digital receiver approach through SBC integration. Performance metrics are explicit. In an 8-hour run with a 50 $\delta T_b(\nu)$1 termination, calibrated, flagged, and averaged spectra show residual RMS of approximately $\delta T_b(\nu)$2 at $\delta T_b(\nu)$3 resolution after a Maximally Smooth fit, with standardized residuals following a Gaussian distribution. Over 44 hours of effective data, residual RMS after Maximally Smooth fitting is $\delta T_b(\nu)$4 at the native resolution; after boxcar averaging by 20 channels, corresponding to effective $\delta T_b(\nu)$5, the RMS is approximately $\delta T_b(\nu)$6. The same paper identifies self-RFI measured inside a non-shielded enclosure and motivates a custom RF enclosure with a target of more than 120 dB isolation.

A later precursor study, STARFIRE-2, uses PRATUSH as the reference radiometer for an Earth-orbit feasibility assessment, denoted PRATUSH-I. In those simulations, PRATUSH-I operates over 55–110 MHz with channel width $\delta T_b(\nu)$7, adopts the PRATUSH antenna beam and $\delta T_b(\nu)$8, and flags channels with brightness temperature greater than $\delta T_b(\nu)$9 as RFI. The study compares near-polar Low-Earth Orbit, near-equatorial Low-Earth Orbit, and GEO, and defines low-RFI zones by a figure of merit with maximum 55, with pixels satisfying $\alpha$0 designated low-RFI. The numerical conclusion is strongly orbit dependent: nighttime low-RFI time per year is $\alpha$1 for polar LEO under isotropic transmitter beams, $\alpha$2 for the same orbit under $\alpha$3 transmitter beams, $\alpha$4 for equatorial LEO, and $\alpha$5 for GEO (Pranesh et al., 21 May 2026).

The Earth-orbit study further reports model-recovery fractions across the Cohen et al. ensemble. In no-RFI and low-RFI Antarctic pixels, 34.6% of models are recovered at $\alpha$6–$\alpha$7, 45.6% at $\alpha$8–$\alpha$9, and 19.7% at $z(\nu) = \frac{1420.40575\ \mathrm{MHz}}{\nu} - 1.$0. In a high-RFI continental pixel, the corresponding fractions are 90.9%, 8.3%, and 0.76%. The paper therefore identifies near-polar LEO as a viable and logistically simpler precursor to a lunar-farside PRATUSH mission, while also noting that L2 without farside shielding still yields Earth-sourced FM RFI of a few $z(\nu) = \frac{1420.40575\ \mathrm{MHz}}{\nu} - 1.$1 up to approximately $z(\nu) = \frac{1420.40575\ \mathrm{MHz}}{\nu} - 1.$2 across 88–110 MHz.

7. Relation to other 21-cm missions and development status

Within the broader landscape of low-frequency space cosmology, PRATUSH is listed among pioneering space-based missions targeting the redshifted 21-cm signal, alongside NCLE, DAPPER, FARSIDE, and DSL. In the white-paper context, PRATUSH is identified as an Indian lunar orbiter mission concept associated with the SARAS radiometer program, and the paper notes that ISRO has provided the SARAS team pre-project funding for development of a lunar orbiter mission—PRATUSH. That source does not provide PRATUSH-specific frequency coverage, bandwidth, collecting area, filling factor, field-of-view, observation mode, or mission lifetime; accordingly, PRATUSH is situated there as a SARAS-led lunar-orbit global 21-cm radiometer concept motivated by the need to escape terrestrial RFI and ionospheric corruption (Koopmans et al., 2019).

The comparison with other missions clarifies PRATUSH’s distinct design choices. Relative to ground-based systems such as EDGES, SARAS, REACH, LEDA, and CTP, PRATUSH emphasizes lunar-farside RFI shielding, especially in the FM band, absence of an ionosphere, absence of Earth ground and horizon coupling, and a designed-in spectrally smooth instrument with in-situ VNA. Relative to other lunar concepts, it chooses an orbiter rather than a lander, accepting a duty cycle in exchange for avoiding ground or regolith coupling. The baseline design further identifies several unique aspects: a dedicated single-payload spacecraft to control EMI, explicit in-situ VNA for impedance tracking, aggressive shielding targets of $z(\nu) = \frac{1420.40575\ \mathrm{MHz}}{\nu} - 1.$3, focus on 55–110 MHz Cosmic Dawn including the FM band, and formal use of Maximally Smooth functions to set instrument smoothness requirements. The characterization “not ‘space-qualified SARAS,’ but shares concepts like double differencing and MS foreground modeling” is therefore precise: PRATUSH inherits calibration philosophy and smooth-antenna design principles from SARAS, without being a direct space translation of the ground system (Rao et al., 8 Jul 2025).

The development path is stated as concept model, then engineering model, then flight model, subject to mission approval. A concept model is under development. Near-term milestones enabled by that model include demonstrating reductions in power, mass, and volume compared to SARAS; exposing electro-mechanical risks with the antenna-over-bus configuration; validating forward models against measurements; building preferred parts lists and identifying low-TRL items; identifying in-situ EMI leakage paths to refine shielding; and characterizing receiver noise while optimizing calibration cadence. Taken together, these milestones indicate that PRATUSH is presently defined as much by its systematics program as by its science case: the mission’s central technical proposition is that a smooth, stable, absolutely calibrated radiometer, operating in the radio-quiet lunar farside environment, can isolate a milliKelvin-level global 21-cm signature against foregrounds at the $z(\nu) = \frac{1420.40575\ \mathrm{MHz}}{\nu} - 1.$4–$z(\nu) = \frac{1420.40575\ \mathrm{MHz}}{\nu} - 1.$5 level.