Joint Sensing & PNT Systems

• JSAP is an integration paradigm that jointly employs waveforms and hardware for simultaneous environment sensing and PNT, optimizing resource use.
• It repurposes communication pilots like the 5G PRS to enable radar sensing alongside positioning, reducing overhead while ensuring accuracy in delay, Doppler, and angle estimation.
• JSAP architectures span terrestrial and non-terrestrial systems, combining structured transforms and parametric estimation to meet rigorous performance and CRLB-based accuracy bounds.

Joint sensing and PNT (JSAP) denotes the integrated use of shared signals, hardware, and processing to perform sensing together with positioning, navigation, and timing. Within the unified taxonomy of multi-functional satellite systems, JSAP appears alongside joint communications and sensing (JCAS), joint communications and PNT (JCAP), and fully integrated joint communications, sensing, and PNT (JCSAP) systems (Sheemar et al., 30 Sep 2025). In terrestrial 5G/6G settings, one concrete instantiation is the use of the 5G Positioning Reference Signal (PRS) as a sensing reference signal, so that the same reference can support radar sensing, communication, and positioning in a single framework (Wei et al., 2022).

1. JSAP as an integration paradigm

JSAP is characterized by joint use of a waveform or payload for two functions that are often designed separately: environment sensing and PNT. In the multi-functional satellite systems survey, the JSAP sub-framework is described as a single satellite carrying a joint-use payload for active Earth-observation and time-of-arrival based PNT, with a digital processing unit, an RF unit, and a shared phased-array antenna (Sheemar et al., 30 Sep 2025). In the terrestrial literature, the same integrative logic appears in 5G New Radio and OFDM-based systems, where pilot or reference signals originally introduced for communication or positioning are repurposed for sensing.

A central implication is architectural rather than merely algorithmic. In the 5G PRS-based scheme, the PRS is treated as a sensing reference signal while remaining compatible with 3GPP TS 38.211/38.214. The resulting arrangement is explicitly described as simultaneously realizing radar sensing, communication and positioning in a convenient manner (Wei et al., 2022). In the satellite survey, the corresponding system-level motivation is resource sharing and functional synergy within a single payload (Sheemar et al., 30 Sep 2025).

The concept should not be conflated with generic integrated sensing and communication. JSAP is narrower: its core coupling is between sensing and PNT, even when communication remains present in the same signal chain. This distinction matters because the estimands and constraints are different. Sensing emphasizes range, velocity, angle, or scene structure, while PNT emphasizes user or platform state, clock bias, timing stability, and navigational observables. The literature summarized here repeatedly operationalizes JSAP through delay, Doppler, and angular parameters that can be mapped either to targets or to user/platform state.

2. Canonical signal models and architectures

A representative terrestrial realization is the 5G PRS monostatic base-station model. A 5G gNB generates downlink PRS as a length-4096 complex Gold sequence, mapped in a comb-pattern over $N$ subcarriers and $M$ OFDM symbols per slot, with comb size $K_{\rm comb}\in\{2,4,6,12\}$. In radar mode, the BS is treated as a monostatic radar observing a point target at range $R_r$ and radial velocity $v$, with round-trip delay and Doppler given by

$\tau=\frac{2R_r}{c}, \qquad f_d=\frac{2v f_c}{c}.$

After dividing the received symbols by the known transmitted PRS, the processed observation becomes

$$S_g[k,m]=\xi e^{-j2\pi K_{\rm comb}k\Delta f\tau} e^{+j2\pi K_{\rm comb}mT_s f_d}+\text{noise},$$

which directly exposes the delay and Doppler structure needed for JSAP estimation (Wei et al., 2022).

A broader 5G NR system model uses a multibeam MIMO base station that simultaneously transmits to a user equipment and scans $N_{\rm dir}$ directions for radar sensing by splitting its total power between a fixed communication beam and a scanning sensing beam. The transmit precoder is

$$\mathbf{w}_T=\sqrt{\alpha}\,\mathbf{w}_{T,s}+\sqrt{1-\alpha}\,\mathbf{w}_{T,c}, \qquad 0\le \alpha\le 1,$$

where $\alpha$ is the sensing-power fraction. In this formulation, the received OFDM echo is a superposition of delayed, Doppler-shifted, angle-weighted copies of the transmitted signal, and target position is then obtained via range and DoA estimation (Pucci et al., 2022).

Uplink JSAP formulations generalize the same structure to parametric channel estimation. For an uplink OFDM pilot block of $M$0 subcarriers and $M$1 OFDM symbols, the base station with $M$2 receive antennas observes

$M$3

with unknown delays $M$4, Dopplers $M$5, angles of departure $M$6, angles of arrival $M$7, and complex gains $M$8. This formulation makes the channel itself the JSAP observation space, since the estimated multipath components can be interpreted geometrically for sensing and PNT (Pinto et al., 2024).

Non-terrestrial JSAP extends these ideas to shared sensing and PNT payloads and to joint satellite–terrestrial systems. In the integrated satellite–terrestrial maritime system, a shore-based terrestrial base station and a low-Earth-orbit satellite provide communication services while simultaneously performing target sensing via reflected echoes of the same downlink signals. The sensing model is bistatic at both nodes, with angle-dependent array responses and CRLB-based localization constraints (Xiong et al., 13 Jul 2026). In the multi-functional satellite systems survey, the abstracted JSAP model uses a continuous-time transmit signal $M$9, a radar sensing channel $K_{\rm comb}\in\{2,4,6,12\}$0, and a PNT observation vector $K_{\rm comb}\in\{2,4,6,12\}$1, thereby placing active sensing and TOA-based positioning in a single signal-theoretic framework (Sheemar et al., 30 Sep 2025).

3. Estimation mechanisms and parameter-to-state mapping

The 5G PRS sensing pipeline is a direct delay–Doppler estimator. Range is recovered by an $K_{\rm comb}\in\{2,4,6,12\}$2-point IFFT across PRS subcarriers,

$K_{\rm comb}\in\{2,4,6,12\}$3

and the peak index yields the range estimate

$K_{\rm comb}\in\{2,4,6,12\}$4

Velocity is recovered by an $K_{\rm comb}\in\{2,4,6,12\}$5-point FFT across symbols,

$K_{\rm comb}\in\{2,4,6,12\}$6

followed by

$K_{\rm comb}\in\{2,4,6,12\}$7

This gives the canonical JSAP pattern in which a communication or positioning reference is transformed into sensing observables through Fourier processing (Wei et al., 2022).

The multibeam 5G NR system uses a 2-D periodogram for range and Doppler,

$K_{\rm comb}\in\{2,4,6,12\}$8

with estimates

$K_{\rm comb}\in\{2,4,6,12\}$9

and therefore

$R_r$0

DoA is estimated by MUSIC after source-number estimation via MDL and covariance construction. In this formulation, target position is obtained from polar-to-Cartesian conversion after range and angle estimation (Pucci et al., 2022).

The uplink parametric approach replaces grid-based peak search with sequential ML/MAP estimation of the physical channel parameters. The algorithm cycles through paths in a “snake” order $R_r$1, and within each path updates $R_r$2 by alternating exact coordinate descent. Except for $R_r$3, each derivative takes the form of a finite Fourier series in the unknown, and the roots can be found by constructing a companion-matrix eigenproblem. After each exact update, a small over-relaxation or momentum step is applied. The significance for JSAP is explicit: time-of-flight $R_r$4 yields a range $R_r$5, AoD and AoA define bearing lines, and Doppler gives relative radial velocity $R_r$6, so the estimator outputs can be fed immediately into joint sensing and PNT engines (Pinto et al., 2024).

A plausible implication is that JSAP algorithm design spans two broad regimes. One regime uses structured transforms such as IFFT/FFT or periodograms when the waveform and geometry are sufficiently simple. The other uses parametric model fitting when the application requires explicit recovery of multipath geometry, angular structure, or temporal priors. The supplied literature supports both regimes.

4. Accuracy bounds, resolutions, and performance metrics

The PRS-based formulation provides explicit resolution, ambiguity, and Cramér–Rao expressions. Its range resolution and unambiguous range are

$R_r$7

and its velocity resolution and unambiguous velocity are

$R_r$8

For radar sensing, when $R_r$9, $v$0, and AWGN SNR, the paper derives closed-form CRLBs: $v$1

$v$2

For positioning with a single OFDM symbol, the ranging CRLB is

$v$3

These expressions are important because they place sensing and positioning performance within a single pilot-based framework (Wei et al., 2022).

The 5G NR multibeam analysis uses a related but system-level metric set: probability of detection, RMSE of position and velocity estimation, and OSPA for multiple targets. Under high SNR and large $v$4, the approximate CRLBs are

$v$5

which propagate to range and velocity. For angle, the ULA CRLB is

$v$6

while beam scanning imposes a floor

$v$7

The same work states that DoA estimation is the dominant error source and that design of narrow scans and high-gain arrays is critical for meter-level PNT (Pucci et al., 2022).

The satellite JSAP survey frames the same issue at a higher level. For positioning, with Gaussian TOA observations, the Fisher Information Matrix is

$v$8

with Position Error Bound given by $v$9. For sensing, the range CRLB is approximated by

$\tau=\frac{2R_r}{c}, \qquad f_d=\frac{2v f_c}{c}.$0

yielding the classical range resolution $\tau=\frac{2R_r}{c}, \qquad f_d=\frac{2v f_c}{c}.$1. This treatment makes explicit that JSAP performance is constrained jointly by bandwidth, observation time, SNR, and synchronization quality (Sheemar et al., 30 Sep 2025).

5. Representative performance results across terrestrial and non-terrestrial settings

The 5G PRS sensing study reports that range RMSE versus SNR shows PRS significantly outperforms SS and DMRS pilots due to its greater $\tau=\frac{2R_r}{c}, \qquad f_d=\frac{2v f_c}{c}.$2 and longer sequence length. It further reports that fractional-Fourier enhancement $\tau=\frac{2R_r}{c}, \qquad f_d=\frac{2v f_c}{c}.$3 reduces range RMSE from meter to decimeter/cm level and velocity RMSE to cm/s level, and that both $\tau=\frac{2R_r}{c}, \qquad f_d=\frac{2v f_c}{c}.$4 and $\tau=\frac{2R_r}{c}, \qquad f_d=\frac{2v f_c}{c}.$5 approach their CRLBs at high SNR. For multi-frame Doppler at $\tau=\frac{2R_r}{c}, \qquad f_d=\frac{2v f_c}{c}.$6 dB, using $\tau=\frac{2R_r}{c}, \qquad f_d=\frac{2v f_c}{c}.$7 frames $\tau=\frac{2R_r}{c}, \qquad f_d=\frac{2v f_c}{c}.$8 ms$\tau=\frac{2R_r}{c}, \qquad f_d=\frac{2v f_c}{c}.$9 and $$S_g[k,m]=\xi e^{-j2\pi K_{\rm comb}k\Delta f\tau} e^{+j2\pi K_{\rm comb}mT_s f_d}+\text{noise},$$0 symbols/frame yields sub-m/s accuracy with only 22.9% per-frame PRS overhead (Wei et al., 2022).

The 5G NR multibeam analysis provides system-level numbers for LOS conditions. At 28 GHz, $$S_g[k,m]=\xi e^{-j2\pi K_{\rm comb}k\Delta f\tau} e^{+j2\pi K_{\rm comb}mT_s f_d}+\text{noise},$$1, $$S_g[k,m]=\xi e^{-j2\pi K_{\rm comb}k\Delta f\tau} e^{+j2\pi K_{\rm comb}mT_s f_d}+\text{noise},$$2, and $$S_g[k,m]=\xi e^{-j2\pi K_{\rm comb}k\Delta f\tau} e^{+j2\pi K_{\rm comb}mT_s f_d}+\text{noise},$$3, for $$S_g[k,m]=\xi e^{-j2\pi K_{\rm comb}k\Delta f\tau} e^{+j2\pi K_{\rm comb}mT_s f_d}+\text{noise},$$4 dB it reports $$S_g[k,m]=\xi e^{-j2\pi K_{\rm comb}k\Delta f\tau} e^{+j2\pi K_{\rm comb}mT_s f_d}+\text{noise},$$5 m, $$S_g[k,m]=\xi e^{-j2\pi K_{\rm comb}k\Delta f\tau} e^{+j2\pi K_{\rm comb}mT_s f_d}+\text{noise},$$6 m/s, $$S_g[k,m]=\xi e^{-j2\pi K_{\rm comb}k\Delta f\tau} e^{+j2\pi K_{\rm comb}mT_s f_d}+\text{noise},$$7, and position error $$S_g[k,m]=\xi e^{-j2\pi K_{\rm comb}k\Delta f\tau} e^{+j2\pi K_{\rm comb}mT_s f_d}+\text{noise},$$8 m. At sub-6 GHz with $$S_g[k,m]=\xi e^{-j2\pi K_{\rm comb}k\Delta f\tau} e^{+j2\pi K_{\rm comb}mT_s f_d}+\text{noise},$$9, at $N_{\rm dir}$0 dB it reports $N_{\rm dir}$1 m, $N_{\rm dir}$2, and position error $N_{\rm dir}$3 m. For detection probability with $N_{\rm dir}$4, the same study reports $N_{\rm dir}$5 for $N_{\rm dir}$6 dB with $N_{\rm dir}$7 or $N_{\rm dir}$8 dB with $N_{\rm dir}$9. In multi-target scenarios, the mean OSPA localization error drops to $$\mathbf{w}_T=\sqrt{\alpha}\,\mathbf{w}_{T,s}+\sqrt{1-\alpha}\,\mathbf{w}_{T,c}, \qquad 0\le \alpha\le 1,$$0 m at $$\mathbf{w}_T=\sqrt{\alpha}\,\mathbf{w}_{T,s}+\sqrt{1-\alpha}\,\mathbf{w}_{T,c}, \qquad 0\le \alpha\le 1,$$1 (Pucci et al., 2022).

The uplink sequential MAP estimator reports detection precision $$\mathbf{w}_T=\sqrt{\alpha}\,\mathbf{w}_{T,s}+\sqrt{1-\alpha}\,\mathbf{w}_{T,c}, \qquad 0\le \alpha\le 1,$$2 and recall $$\mathbf{w}_T=\sqrt{\alpha}\,\mathbf{w}_{T,s}+\sqrt{1-\alpha}\,\mathbf{w}_{T,c}, \qquad 0\le \alpha\le 1,$$3 for up to $$\mathbf{w}_T=\sqrt{\alpha}\,\mathbf{w}_{T,s}+\sqrt{1-\alpha}\,\mathbf{w}_{T,c}, \qquad 0\le \alpha\le 1,$$4 paths in a 60 GHz indoor setting, using 50 OFDM symbols $$\mathbf{w}_T=\sqrt{\alpha}\,\mathbf{w}_{T,s}+\sqrt{1-\alpha}\,\mathbf{w}_{T,c}, \qquad 0\le \alpha\le 1,$$5 40 subcarriers. It also reports parameter estimation MSEs on the order of $$\mathbf{w}_T=\sqrt{\alpha}\,\mathbf{w}_{T,s}+\sqrt{1-\alpha}\,\mathbf{w}_{T,c}, \qquad 0\le \alpha\le 1,$$6 s$$\mathbf{w}_T=\sqrt{\alpha}\,\mathbf{w}_{T,s}+\sqrt{1-\alpha}\,\mathbf{w}_{T,c}, \qquad 0\le \alpha\le 1,$$7, $$\mathbf{w}_T=\sqrt{\alpha}\,\mathbf{w}_{T,s}+\sqrt{1-\alpha}\,\mathbf{w}_{T,c}, \qquad 0\le \alpha\le 1,$$8 rad$$\mathbf{w}_T=\sqrt{\alpha}\,\mathbf{w}_{T,s}+\sqrt{1-\alpha}\,\mathbf{w}_{T,c}, \qquad 0\le \alpha\le 1,$$9, and $\alpha$0 Hz$\alpha$1, together with map reconstructions whose median reflector localization error was under 1 m. In complexity terms, it states that overall complexity grows roughly as $\alpha$2, with typical single-core run-times on the order of 10–20 ms for $\alpha$3, $\alpha$4, $\alpha$5 (Pinto et al., 2024).

The integrated satellite–terrestrial maritime study extends the performance discussion to beamforming-constrained non-terrestrial JSAP. It reports DE localization convergence in 30–40 generations and beamforming SCA-SDP convergence in 10–15 iterations. Under tighter CRB constraints, sum-rate drops, while larger arrays $\alpha$6 yield $\alpha$7 rate gain for the same $\alpha$8. It also reports that dual-function signals outperform separate communication-only “single-function” beams by 10–15%, that pure-communication ZFBF and Lagrange-dual beamforming achieve up to 25% lower rates once sensing constraints bind, and that two-node joint DE achieves sub-meter RMSE at modest complexity (Xiong et al., 13 Jul 2026).

6. Design tensions, misconceptions, and open research questions

A recurring design variable in terrestrial JSAP is the trade-off between sensing accuracy, refresh time, and communication overhead. In the PRS framework, velocity resolution under multi-frame accumulation becomes

$\alpha$9

which improves as $M$00 grows. The same source states the trade-offs explicitly: increasing $M$01 or $M$02 reduces $M$03 but increases either sensing refresh time $M$04 or frame overhead $M$05. It therefore recommends a dedicated “sensing subframe” in each 10 ms frame, flexible comb patterns, symbol counts, and power boosts of sensing reference signals, waveform reconfiguration $M$06 on the fly, multi-cell comb orthogonalization, joint use of data symbols and sensing pilots, and hybrid OFDM-chirp pilots (Wei et al., 2022).

Power allocation creates an analogous tension in multibeam 5G NR and in satellite–terrestrial JSAP. In the 5G NR study, reducing the sensing-power fraction $M$07 preserves UE throughput but degrades PNT accuracy roughly proportionally. At 3.5 GHz with $M$08, with $M$09 the localization RMSE stays $M$10 m out to 250 m, whereas with $M$11 it is $M$12 m; at 28 GHz with $M$13, with $M$14 RMSE is $M$15 m to 85 m, whereas with $M$16 RMSE is $M$17 m (Pucci et al., 2022). In the maritime system, the joint beamforming problem is posed directly as sum-rate maximization subject to power and localization-accuracy constraints, making the rate–accuracy trade-off explicit (Xiong et al., 13 Jul 2026).

Several misconceptions are addressed implicitly by the literature. One is that communication or positioning pilots are unsuitable for sensing. The PRS study argues the opposite by showing feasibility and superiority of PRS over other pilot signals in radar sensing, and by concluding that PRS can serve simultaneously as a PNT ranging pilot, a radar sensing reference for range and Doppler, and a communications channel estimation pilot with no hardware changes (Wei et al., 2022). Another is that multipath is uniformly harmful. The satellite survey states instead that multipath aids PNT but distorts radar images; balancing these effects is nontrivial (Sheemar et al., 30 Sep 2025). This suggests that JSAP receiver design must often decide whether a path is a nuisance, a geometric cue, or both.

Open problems are stated most systematically in the satellite survey. The listed challenges are accuracy, coverage, spectrum, coordination, transmit power, and multipath trade-offs. The listed future research directions are dual-utility signal design, wideband/multi-band architectures, sub-ns calibration and synchronization, low-power enhancement algorithms, super-resolution and compressed-sensing processing, context-adaptive interference rejection, benchmarks and testbeds, and AI-driven context awareness (Sheemar et al., 30 Sep 2025). A plausible implication is that future JSAP systems will be judged less by whether they can share a waveform than by whether they can maintain CRLB-limited estimation, synchronization integrity, and resource efficiency under realistic multi-node, multi-band, and multi-path operating conditions.