Narrowband Sweeper Receiver (NSR)

• NSR is a family of receivers that acquire narrowband power measurements via frequency sweeps, enabling precise spectral reconstruction.
• It supports diverse applications, including SAR antenna-pattern estimation, RSS-based occupancy sensing, and drifting-tone searches in technosignature detection.
• The design leverages frequency diversity to enhance SNR while trading off coherent phase recovery, making calibration and local linearity crucial.

Searching arXiv for the cited papers to ground the article in current records. A Narrowband Sweeper Receiver (NSR) denotes, in the literature considered here, a receiver class that acquires narrowband or frequency-resolved measurements across a swept band, across multiple discrete channels, or across hypothesized drift trajectories, and then performs inference from the resulting frequency structure rather than from wideband delay resolution. The term is used explicitly for in-orbit SAR antenna-pattern estimation, where a swept narrowband front end measures power frequency-by-frequency within a broadband chirp, and it is also used implicitly or operationally in swept-channel RSS sensing and narrowband drift-search pipelines for technosignature detection. This suggests that NSR is best treated as a technical family of frequency-swept or frequency-resolved receivers rather than as a single canonical hardware topology (Roueinfar et al., 31 Jul 2025, Li et al., 19 Mar 2026).

1. Definition and scope

In the most explicit formulation, the NSR is a power-measuring spectral sweeper: the receive chain isolates one narrow spectral slice at a time, measures its power, and reconstructs either a spectrum or a frequency-indexed physical quantity from those measurements. The Cosmo-SkyMed study implements this directly with a receive antenna, low-noise amplifier (LNA), bandpass filter (BPF), power divider, and an NSR branch comprising mixer, swept VCO, narrowband bandpass filter (NBPF), detector logarithmic video amplifier (DLVA), and ADC; the proposed structure is stated to be equivalent to an NBPF moving through the spectrum (Roueinfar et al., 31 Jul 2025).

A broader NSR interpretation appears in applications where the receiver does not sweep a continuously tunable analog filter, but still constructs detection statistics from narrowband measurements indexed by frequency or drift template. In near-line-of-sight occupancy sensing, RSS samples are taken over a set of narrowband channels and processed as a swept-frequency signature. In technosignature search, an offline de-Doppler bank operates as a software narrowband sweeper over linear drift-rate templates. By contrast, the PLC FRESH-filter receiver does not define or use the term Narrowband Sweeper Receiver; it is explicitly a cyclostationary frequency-shift filtering receiver rather than a sweep/scanning receiver (Shlezinger et al., 2014).

The distinction from other narrowband receivers is therefore operational. An NSR is not defined merely by narrow bandwidth, but by the use of narrow spectral selectivity as a sensing resource. It may be power-only rather than coherent, as in the DLVA-based SAR receiver, or RSS-only rather than phase-preserving, as in swept-channel human-reflection detection (Roueinfar et al., 31 Jul 2025, Yiğitler et al., 2014).

2. Swept-channel RSS sensing and human-induced reflections

A particularly clear NSR sensing regime is the detection of a human near, but not blocking, the line-of-sight between a transmitter and receiver using RSS measurements across multiple narrowband channels. The geometric core is the excess reflected path length

$$\Delta = \left\| \boldsymbol{p} - \boldsymbol{p}_t\right\| + \left\| \boldsymbol{p} - \boldsymbol{p}_r\right\| - d,$$

which induces a frequency-dependent phase shift

$\phi_l = 2\pi \Delta \frac{f_{c,l}}{c_0}.$

Because the direct and reflected paths interfere, the reflection contribution in dB becomes periodic in communication frequency, and the swept-channel RSS vector constitutes a deterministic signature of near-LoS human presence (Yiğitler et al., 2014).

The received power model is written as

$$\mathcal{P}_r^{dB} = \mathcal{P}_{LoS}^{dB}(f_c) + \zeta(\Delta,\Gamma,f_c) + \nu(f_c),$$

with baseline subtraction yielding

$$z(\Delta,\Gamma,f_c) \triangleq \zeta(\Delta,\Gamma,f_c) + \nu(f_c).$$

The paper then derives a Fourier-series representation of the reflection term,

$$\zeta(\Delta,\Gamma,\beta) = -2\hat{e}\sum_{i=1}^{\infty} a_i \cos(2\pi i \Delta \beta),$$

showing that the frequency of the RSS ripple is set by $\Delta$, while the amplitude is controlled by $A=\Gamma(1+\Delta/d)^{-\eta/2}$. For $\Gamma \le 0.7$, the first two harmonics carry more than $96.76\%$ of the total signal power, which motivates a low-complexity two-harmonic approximation (Yiğitler et al., 2014).

Detection is performed by energy accumulation across swept channels,

$\mathcal{E}_z = \sum_{l=1}^{C} z_l^2,$

with a Neyman-Pearson threshold

$\phi_l = 2\pi \Delta \frac{f_{c,l}}{c_0}.$0

Under the stated assumptions, $\phi_l = 2\pi \Delta \frac{f_{c,l}}{c_0}.$1 is central chi-square under $\phi_l = 2\pi \Delta \frac{f_{c,l}}{c_0}.$2 and noncentral chi-square under $\phi_l = 2\pi \Delta \frac{f_{c,l}}{c_0}.$3. The principal engineering result is that increasing the number of channels improves performance both because the frequency-periodic reflection is sampled more richly and because signal-noise cross terms average out more effectively. Experimentally, the abstract states that with more than eight frequency channels, a single TX-RX pair can detect a person with detection probability higher than $\phi_l = 2\pi \Delta \frac{f_{c,l}}{c_0}.$4 and false alarm probability less than $\phi_l = 2\pi \Delta \frac{f_{c,l}}{c_0}.$5 in an area of $\phi_l = 2\pi \Delta \frac{f_{c,l}}{c_0}.$6 m$\phi_l = 2\pi \Delta \frac{f_{c,l}}{c_0}.$7 (Yiğitler et al., 2014).

This use of RSS-only swept-frequency diversity is notable because the receiver does not require output phase measurements or channel impulse response reconstruction. The method is, however, explicitly tied to a near-LoS, single-bounce reflection model, baseline calibration, and a stationary or slowly varying environment.

3. Drift-search NSR in technosignature detection

In SETI and related technosignature searches, the NSR concept appears as a drifting-tone receiver that searches for narrowband carriers whose apparent frequency evolves in time. For Earth–exoplanet geometry, the long-term physical model gives

$\phi_l = 2\pi \Delta \frac{f_{c,l}}{c_0}.$8

with the total radial acceleration decomposed into four dominant celestial-mechanics terms: Earth rotation, exoplanet rotation, Earth orbit, and exoplanet orbit. Over short observations, the signal frequency can usually be treated as drifting linearly with approximately constant drift rate. Over long observations, the drift is not generally linear and should follow an intermittent pseudosinusoidal trajectory; higher orbital eccentricity introduces asymmetric drifting (Li et al., 2022).

This distinction directly constrains NSR architecture. A constant-drift deDoppler bank is appropriate for short coherent windows, whereas long-term candidate confirmation should test whether detections align with periodic or pseudosinusoidal drift evolution. The same paper gives the minimum distinguishable drift rate as

$\phi_l = 2\pi \Delta \frac{f_{c,l}}{c_0}.$9

making explicit the coupling between frequency resolution and observation duration (Li et al., 2022).

A practical implementation appears in the FAST search toward 3I/ATLAS. The observations used the FAST L-band 19-beam multibeam receiver, with raw coverage from $$\mathcal{P}_r^{dB} = \mathcal{P}_{LoS}^{dB}(f_c) + \zeta(\Delta,\Gamma,f_c) + \nu(f_c),$$0–$$\mathcal{P}_r^{dB} = \mathcal{P}_{LoS}^{dB}(f_c) + \zeta(\Delta,\Gamma,f_c) + \nu(f_c),$$1 GHz, search restricted to $$\mathcal{P}_r^{dB} = \mathcal{P}_{LoS}^{dB}(f_c) + \zeta(\Delta,\Gamma,f_c) + \nu(f_c),$$2–$$\mathcal{P}_r^{dB} = \mathcal{P}_{LoS}^{dB}(f_c) + \zeta(\Delta,\Gamma,f_c) + \nu(f_c),$$3 GHz, sampling time $$\mathcal{P}_r^{dB} = \mathcal{P}_{LoS}^{dB}(f_c) + \zeta(\Delta,\Gamma,f_c) + \nu(f_c),$$4, and frequency resolution $$\mathcal{P}_r^{dB} = \mathcal{P}_{LoS}^{dB}(f_c) + \zeta(\Delta,\Gamma,f_c) + \nu(f_c),$$5 Hz. The narrowband drift search was carried out with signal-to-noise ratio over 10 via the bliss pipeline over $$\mathcal{P}_r^{dB} = \mathcal{P}_{LoS}^{dB}(f_c) + \zeta(\Delta,\Gamma,f_c) + \nu(f_c),$$6, with hits grouped into events and then filtered by cluster analysis and drift-rate cut-off. The multibeam logic uses the central beam as on-source and beams $$\mathcal{P}_r^{dB} = \mathcal{P}_{LoS}^{dB}(f_c) + \zeta(\Delta,\Gamma,f_c) + \nu(f_c),$$7 as simultaneous off-source references, so that spatial coincidence functions as a real-time veto against many RFI classes (Li et al., 19 Mar 2026).

The formal hit definition is

$$\mathcal{P}_r^{dB} = \mathcal{P}_{LoS}^{dB}(f_c) + \zeta(\Delta,\Gamma,f_c) + \nu(f_c),$$8

and the paper supplements conventional de-drift thresholding with HDBSCAN clustering, a multibeam coincidence test, structure-tensor morphology, PCA-based characterization, and visual inspection. No credible narrowband radio technosignature was detected; the null results place constraints on transmitters above $$\mathcal{P}_r^{dB} = \mathcal{P}_{LoS}^{dB}(f_c) + \zeta(\Delta,\Gamma,f_c) + \nu(f_c),$$9 W (Li et al., 19 Mar 2026).

Taken together, these works define a two-timescale NSR logic for drifting tones: linear drift is the correct local model, while pseudosinusoidal evolution is the correct long-term physical model.

4. In-orbit SAR antenna-pattern estimation

The most direct and hardware-specific use of the term NSR is the in-orbit Cosmo-SkyMed antenna-pattern study. Its motivation is that a spaceborne SAR transmitter emits a broadband pulsed LFM chirp, but the antenna pattern of an active phased array is frequency dependent. If a conventional simple envelope detector (SED) measures only total power over the full occupied bandwidth, then the extracted pattern is frequency-integrated, so null positions shift with frequency, sidelobe structure changes with frequency, and averaging over the chirp blurs these features (Roueinfar et al., 31 Jul 2025).

The NSR avoids that bias by sweeping a narrow receive window across the SAR spectrum and estimating power frequency-by-frequency. The array analysis isolates the array factor,

$$z(\Delta,\Gamma,f_c) \triangleq \zeta(\Delta,\Gamma,f_c) + \nu(f_c).$$0

and the paper states that null spacing increases proportionally with wavelength. This provides the theoretical reason that broadband integration is not an antenna-pattern measurement at any single frequency (Roueinfar et al., 31 Jul 2025).

Operationally, the key sweep timing relation is

$$z(\Delta,\Gamma,f_c) \triangleq \zeta(\Delta,\Gamma,f_c) + \nu(f_c).$$1

together with the pulse-compatibility requirement

$$z(\Delta,\Gamma,f_c) \triangleq \zeta(\Delta,\Gamma,f_c) + \nu(f_c).$$2

The paper reports bad reconstruction for $$z(\Delta,\Gamma,f_c) \triangleq \zeta(\Delta,\Gamma,f_c) + \nu(f_c).$$3 and good reconstruction for $$z(\Delta,\Gamma,f_c) \triangleq \zeta(\Delta,\Gamma,f_c) + \nu(f_c).$$4. This establishes a central NSR constraint for pulsed waveforms: step duration must be compatible with the pulse train (Roueinfar et al., 31 Jul 2025).

The SNR rationale is equally explicit. Using the standard $$z(\Delta,\Gamma,f_c) \triangleq \zeta(\Delta,\Gamma,f_c) + \nu(f_c).$$5 dependence, the study gives

$$z(\Delta,\Gamma,f_c) \triangleq \zeta(\Delta,\Gamma,f_c) + \nu(f_c).$$6

For the implemented case, the SED bandwidth is $$z(\Delta,\Gamma,f_c) \triangleq \zeta(\Delta,\Gamma,f_c) + \nu(f_c).$$7 and the NSR bandwidth is $$z(\Delta,\Gamma,f_c) \triangleq \zeta(\Delta,\Gamma,f_c) + \nu(f_c).$$8, yielding a theoretical SNR gain of

$$z(\Delta,\Gamma,f_c) \triangleq \zeta(\Delta,\Gamma,f_c) + \nu(f_c).$$9

or roughly $$\zeta(\Delta,\Gamma,\beta) = -2\hat{e}\sum_{i=1}^{\infty} a_i \cos(2\pi i \Delta \beta),$$0. The in-orbit experiment received a Cosmo-SkyMed signal for $$\zeta(\Delta,\Gamma,\beta) = -2\hat{e}\sum_{i=1}^{\infty} a_i \cos(2\pi i \Delta \beta),$$1 seconds, with NSR sweep time $$\zeta(\Delta,\Gamma,\beta) = -2\hat{e}\sum_{i=1}^{\infty} a_i \cos(2\pi i \Delta \beta),$$2, giving $$\zeta(\Delta,\Gamma,\beta) = -2\hat{e}\sum_{i=1}^{\infty} a_i \cos(2\pi i \Delta \beta),$$3 points per second at each frequency. Pattern slices at $$\zeta(\Delta,\Gamma,\beta) = -2\hat{e}\sum_{i=1}^{\infty} a_i \cos(2\pi i \Delta \beta),$$4 MHz and $$\zeta(\Delta,\Gamma,\beta) = -2\hat{e}\sum_{i=1}^{\infty} a_i \cos(2\pi i \Delta \beta),$$5 MHz differed, with larger null spacing at the lower frequency, and the NSR-derived pattern was reported to be sharper and more faithful than the SED estimate (Roueinfar et al., 31 Jul 2025).

This formulation is paradigmatic: the NSR is both a spectral isolator and an SNR enhancer, and its output is frequency-resolved power rather than coherent I/Q.

5. Adjacent architectures and terminological boundaries

Two closely related architectures help delimit what NSR is and is not. The first is the receiver for OFDM recovery in narrowband PLC based on frequency-shift filtering. That paper explicitly states that it does not define or use the term Narrowband Sweeper Receiver. Its receiver is a two-stage serial FRESH-filter architecture that processes selected cyclic frequency translations in parallel to estimate cyclostationary noise and then recover the OFDM signal. It is a fixed multi-branch cyclostationary filter bank matched to known cyclic frequencies, not a sweep/scanning receiver (Shlezinger et al., 2014).

The second is Band-Sweeping M-ary PSK. There, the carrier is frequency modulated by a periodic ramp so that its instantaneous frequency sweeps the allocated transmission band, and the received swept FM carrier becomes amplitude-modulated by the channel transfer function. The receiver then uses envelope detection to estimate the channel magnitude response without dedicated pilot symbols. The main design relations are

$$\zeta(\Delta,\Gamma,\beta) = -2\hat{e}\sum_{i=1}^{\infty} a_i \cos(2\pi i \Delta \beta),$$6

$$\zeta(\Delta,\Gamma,\beta) = -2\hat{e}\sum_{i=1}^{\infty} a_i \cos(2\pi i \Delta \beta),$$7

and

$$\zeta(\Delta,\Gamma,\beta) = -2\hat{e}\sum_{i=1}^{\infty} a_i \cos(2\pi i \Delta \beta),$$8

This is not a conventional stand-alone NSR in the instrumentation sense, but it is a swept-carrier transceiver with an embedded narrowband probing mechanism (Alaa, 2014).

These comparisons clarify a recurrent misconception. A receiver need not be called an NSR merely because it is narrowband, and it need not physically tune a single analog filter to be NSR-like in operation. The defining feature is the deliberate exploitation of narrow instantaneous spectral selectivity across frequency or drift template.

6. Recurring design principles and limitations

Across these works, several design principles recur. First, frequency diversity is the sensing resource. In RSS-based occupancy detection, one should measure RSS quickly over multiple narrowband channels, subtract a calibrated empty-room baseline per channel, and compute a swept-channel energy statistic. In SAR calibration, one should make the receive window narrow enough to suppress frequency-averaging bias while keeping sweep timing compatible with $$\zeta(\Delta,\Gamma,\beta) = -2\hat{e}\sum_{i=1}^{\infty} a_i \cos(2\pi i \Delta \beta),$$9. In drifting-tone search, one should channelize finely enough that the expected drift per time sample remains smaller than or comparable to one frequency bin, preserving dechirping efficiency (Yiğitler et al., 2014, Roueinfar et al., 31 Jul 2025, Li et al., 19 Mar 2026).

Second, calibration is essential. The RSS detector requires a vacant-scene baseline and a variance estimate, and the SAR NSR requires receiver frequency-response equalization and stable mapping from sweep position to RF frequency. The FAST technosignature search further shows that thresholding alone is insufficient in a realistic RF environment: millions of hits can survive an SNR threshold of 10, and multibeam vetoes, clustering, morphology features, and instrument-specific RFI knowledge remain necessary (Yiğitler et al., 2014, Roueinfar et al., 31 Jul 2025, Li et al., 19 Mar 2026).

Third, local linear models are often valid, but global linearity is not. The 3I/ATLAS and exoplanet-drift studies rely on constant drift templates over finite dwells, yet the underlying long-term frequency evolution is intermittent pseudosinusoidal and can be asymmetric for eccentric orbits. Similarly, the human-reflection RSS model is local to a near-LoS, single-bounce regime rather than a general moving-multipath model (Li et al., 2022, Yiğitler et al., 2014).

The principal limitations are equally consistent. Power-only NSRs discard coherent phase information; the Cosmo-SkyMed receiver uses a DLVA rather than complex sampling. RSS-only human-reflection sensing does not directly localize with wideband-range resolution and cannot resolve multiple reflectors in delay. Long-term drift shape is a useful discriminator in technosignature search, but not a standalone proof of extraterrestrial origin, since some RFI can also produce pseudosinusoidal or chirp-like signatures. This suggests that NSR methods are especially effective for occupancy assessment, coarse near-LoS presence sensing, frequency-resolved antenna-pattern extraction, and drifting-tone search, but are less informative than coherent wideband radar or full complex-sampling receivers when delay resolution, phase recovery, or multi-source separation is the primary objective (Roueinfar et al., 31 Jul 2025, Li et al., 2022, Li et al., 19 Mar 2026).