[2512.21937] Integrating Low-Altitude SAR Imaging into UAV Data Backhaul
A Dual-Purpose Approach to Low-Altitude Communications
BLUF (Bottom Line Up Front): Researchers have developed a novel framework that enables unmanned aerial vehicles (UAVs) to simultaneously perform high-resolution synthetic aperture radar (SAR) imaging and cellular data communications using standard 5G OFDM signals, potentially revolutionizing low-altitude wireless networks by eliminating the need for separate sensing and communications hardware while dramatically improving imaging performance over pilot-only approaches. However, the conflicting trajectory requirements between optimal SAR formation and reliable communications present fundamental constraints that require sophisticated path planning and adaptive resource allocation.
The Convergence of Sensing and Communications
The airspace below 3,000 meters is becoming increasingly crowded with unmanned aerial vehicles performing everything from package delivery to infrastructure monitoring. A fundamental challenge has emerged: these UAVs need both to sense their environment through radar imaging and transmit large volumes of data back to ground stations—traditionally requiring separate, heavy hardware systems that drain battery life and limit flight time.
Figure 1 illustrates a breakthrough solution to this problem: a joint OFDM-SAR imaging and data backhaul system. The diagram depicts a UAV platform moving along a flight path (the y-axis) at altitude, simultaneously illuminating ground targets with its radar beam while maintaining communications with cellular base stations arranged in a hexagonal coverage pattern below.
The geometry shows several key elements: the UAV operates at height Hp above ground level, with its radar beam footprint (shown in gray) covering multiple hexagonal cells containing base station towers. Within this footprint, individual scatterers—the targets being imaged—are positioned at coordinates (xq, yq, 0) on the ground plane. The slant range R̄q represents the direct line-of-sight distance from the UAV to each target, while Rq,m captures the instantaneous range that varies as the platform moves through different azimuth positions during its synthetic aperture.
From Separate Systems to Integrated Architecture
Traditional airborne SAR systems transmit deterministic radar waveforms—typically linear frequency modulated chirps—that are optimized purely for imaging. Meanwhile, communications require random data symbols that encode information. Historically, these conflicting requirements meant UAVs carried separate radar and communications modules, significantly increasing payload weight, power consumption, and system complexity.
The integrated sensing and communications (ISAC) paradigm seeks to unify these functions using a single waveform. According to Liu et al. (2022), "Integrating sensing and communications for ubiquitous IoT" represents a fundamental shift toward dual-functional wireless networks. Current 5G New Radio implementations already include sensing capabilities through Sounding Reference Signals (SRS), but these occupy only small fractions of available spectrum—less than 10% in typical configurations—severely limiting imaging resolution and quality.
The research presented in the uploaded paper by Du et al. (2025) takes a radically different "communication-centric" approach: instead of dedicating spectrum to deterministic sensing signals, the system exploits the random data symbols themselves for imaging purposes. This "data-aided imaging" approach uses the full bandwidth and symbol density of the communications waveform, achieving what the pilot-only approach cannot.
The Technical Challenge: Randomness as Friend and Foe
The OFDM (Orthogonal Frequency Division Multiplexing) waveform has become the standard for modern wireless communications due to its spectral efficiency and resistance to multipath interference. It's the foundation of WiFi, 4G LTE, and 5G networks. Its structure—dividing the available bandwidth into thousands of narrowband subcarriers—also makes it attractive for radar, as the multi-carrier architecture naturally provides a wide instantaneous bandwidth critical for range resolution.
However, a fundamental problem arises when attempting to use communications data for imaging: the random nature of data symbols drawn from constellations like 256-QAM (Quadrature Amplitude Modulation) introduces amplitude and phase variations that distort the radar echo structure. Unlike deterministic radar waveforms where the transmitted signal is precisely known, random data symbols create statistical uncertainties that degrade image quality.
The research team led by Du addresses this through sophisticated temporal-frequency (TF) domain filtering schemes applied before standard range-Doppler (RD) imaging algorithms. Three filtering approaches are evaluated:
- Matched Filtering (MF): Maximizes signal-to-noise ratio by correlating with the complex conjugate of transmitted symbols
- Reciprocal Filtering (RF): Performs element-wise division to normalize symbol variations
- Wiener Filtering (WF): Applies linear minimum mean square error optimization, balancing randomness suppression against noise amplification
According to recent ISAC research by Keskin et al. (2025) in "Fundamental tradeoffs in monostatic ISAC," these filtering choices create complex tradeoffs between imaging quality, computational complexity, and required prior knowledge about signal-to-noise ratios.
Geometry and Coverage Analysis
The geometric analysis in Figure 1 addresses a critical practical question: can a UAV maintain continuous communications coverage while performing SAR imaging? The answer depends on the relationship between the radar beam footprint and cellular base station spacing.
For a UAV operating at 1,000 meters altitude with a 3.5 GHz carrier frequency and 0.1-meter antenna apertures, the azimuth coverage length reaches approximately 564 meters while elevation coverage extends to roughly 1,899 meters—both substantially exceeding typical 5G cell radii of 100-500 meters in urban environments. This ensures that as the UAV sweeps along its flight path, its beam footprint encompasses at least one base station at all times, enabling uninterrupted data backhaul even as the platform moves.
The slant range geometry becomes particularly important for signal processing. The range Rq,m between the UAV and each ground point varies quadratically with azimuth position due to the platform's motion. This creates "range cell migration"—the phenomenon where a stationary target's echo appears to shift across different range bins as the UAV moves. Compensating for this migration is essential for achieving focused images, and the uploaded paper demonstrates how the OFDM signal structure with its cyclic prefix naturally mitigates the interference that would otherwise arise from this migration.
The Trajectory Dilemma: Competing Requirements for SAR and Communications
While the waveform integration solves one fundamental challenge, a deeper constraint emerges from the inherent conflict between optimal trajectories for SAR imaging versus reliable communications—a tension that becomes particularly acute in low-altitude wireless networks.
SAR Trajectory Requirements
Classical SAR imaging theory imposes strict constraints on platform motion. The synthetic aperture principle relies on coherently combining echoes from many spatial positions as the platform moves, effectively creating a much larger antenna than physically possible. This coherent processing demands:
Linear trajectory stability: The platform must maintain a straight flight path with minimal deviations. According to Cumming and Wong (2005), trajectory deviations exceeding one-eighth wavelength (approximately 2.1 cm at 3.5 GHz) introduce phase errors that degrade focusing performance. Modern motion compensation techniques can correct for measured deviations, but these require precise inertial measurement units and add computational complexity.
Constant velocity: SAR processing algorithms assume uniform sample spacing in the synthetic aperture domain. Velocity variations cause non-uniform azimuth sampling, leading to geometric distortions in the reconstructed image. For the 50 m/s platform velocity specified in the paper, maintaining stability within ±0.5 m/s is typically necessary for high-quality imaging.
Perpendicular orientation: Stripmap SAR geometry, as depicted in Figure 1, requires the velocity vector to be perpendicular to the radar look direction. This broadside configuration ensures uniform illumination time across all targets in the swath. Deviation from perpendicularity (squint angle) shortens the effective synthetic aperture for off-track targets, degrading azimuth resolution asymmetrically across the image.
Extended dwell time: Azimuth resolution ρa = v/(2Ba) improves with larger azimuth bandwidth Ba = KaTa, which requires longer coherent integration times Ta. The paper specifies Ta = 2 seconds, during which the UAV travels 100 meters along a straight path. Any requirement to deviate from this path—for obstacle avoidance, wind compensation, or communications optimization—directly degrades imaging performance.
Communications Trajectory Requirements
Cellular communications, in contrast, benefit from very different trajectory characteristics:
Base station proximity: Signal strength follows the Friis transmission equation, decreasing with the square of distance in free space. In multipath-rich low-altitude environments, path loss increases even more rapidly. Communications link quality improves dramatically when the UAV maintains proximity to base stations, favoring trajectories that pass near or hover above cellular towers rather than following straight lines between them.
Elevation angle optimization: The antenna radiation patterns of ground base stations are optimized for terrestrial users, with maximum gain at shallow elevation angles (0-30 degrees). As a UAV passes directly overhead, it enters the antenna's null region where gain drops by 10-20 dB. According to Zeng et al. (2019) in "Accessing from the sky: A tutorial on UAV communications for 5G and beyond," maintaining elevation angles between 15-45 degrees optimizes link quality, requiring UAVs to maintain lateral offset from base stations rather than overflying them.
Handover minimization: As UAVs move at 50 m/s, they can traverse a small cell in 10-20 seconds. Each handover between base stations introduces latency, packet loss, and control overhead. Trajectories that minimize cell boundary crossings—perhaps following cell edges or spiraling within cells—reduce handover frequency but conflict with straight SAR flight paths.
Line-of-sight maintenance: Low-altitude urban environments present significant blockage from buildings, trees, and terrain. Communications reliability demands trajectories maintaining clear line-of-sight to serving base stations. The optimal altitude depends on local topography: too low and buildings block signals, too high and the UAV enters antenna null regions. This altitude optimization rarely aligns with the constant-altitude requirement for SAR swath uniformity.
Quantifying the Conflict
The fundamental tension can be expressed mathematically. For SAR imaging, the Cramér-Rao lower bound on achievable resolution sets minimum requirements on trajectory stability. According to Meta et al. (2010) in "FMCW SAR based on a single-chip MIMO radar," phase error standard deviation must satisfy:
σφ < λ/(4πL)
where L is the synthetic aperture length. For L = 100 m and λ = 0.086 m (3.5 GHz), this yields σφ < 0.068 radians or about 4 degrees—an extremely tight constraint.
Meanwhile, communications capacity follows Shannon's theorem, C = B log₂(1 + SNR), where SNR degrades with distance from base stations. The path loss exponent in low-altitude environments typically ranges from 2.5-4.0 depending on clutter density. A UAV deviating 200 meters from optimal base station proximity to maintain SAR trajectory might experience 15-25 dB additional path loss, reducing communications capacity by factors of 30-300.
Recent research by Zhang et al. (2024) in "Joint trajectory and resource optimization for UAV-enabled integrated sensing and communication systems" quantifies these tradeoffs using multi-objective optimization frameworks. Their analysis reveals that naive approaches—simply flying straight lines for SAR while hoping for adequate communications—achieve only 30-40% of the capacity available from communications-optimized trajectories, while communications-optimized paths degrade SAR image quality by similar margins.
Resolution Strategies: From Conflict to Coexistence
Addressing these competing constraints requires sophisticated approaches that transcend simple compromise between extremes. Recent research has explored several promising directions:
1. Segmented Mission Profiles
Rather than attempting to optimize both functions simultaneously throughout the flight, mission profiles can alternate between SAR-optimized and communications-optimized segments. During SAR collection phases, the UAV flies straight, level trajectories optimized for image quality while accepting reduced communications link margins. During communications phases, the platform maneuvers to optimal relay positions, offloading accumulated imaging data while accepting that SAR coherence is temporarily lost.
This approach requires careful buffer management and latency tolerance analysis. The paper specifies SAR integration times of Ta = 2 seconds generating image data that must be transmitted. At 100 MHz bandwidth with 256-QAM modulation, instantaneous data rates can exceed 500 Mbps, but link quality variations during SAR trajectories may reduce effective throughput to 50-100 Mbps. Buffer sizing must accommodate the mismatch between image generation rate and time-averaged communications capacity.
Wu et al. (2025) in "Toward multi-functional LAWNs with ISAC" propose adaptive segmentation where the duty cycle between SAR and communications phases adjusts dynamically based on imaging requirements, communications backlog, and link conditions. For time-critical applications like disaster response, the system might prioritize rapid image transmission even if imaging quality suffers slightly. For mapping missions, image quality takes precedence with communications relegated to non-critical periods.
2. Opportunistic SAR Collection
An alternative paradigm inverts the traditional approach: rather than forcing communications trajectories to accommodate SAR requirements, the system opportunistically collects SAR data whenever trajectory constraints permit. Modern UAV path planning algorithms incorporate obstacle avoidance, wind compensation, and energy optimization. These algorithms naturally produce trajectory segments that approximate the straight-line, constant-velocity conditions SAR requires.
The key innovation is adaptive aperture synthesis. Instead of requiring 2-second coherent integration times along perfectly straight paths, advanced autofocus algorithms can exploit shorter segments—perhaps 0.5-1.0 seconds—of sufficiently linear motion. Recent developments in compressed sensing SAR, reviewed by Ender (2010) in "On compressive sensing applied to radar," demonstrate that high-quality images can be reconstructed from non-uniformly sampled apertures when the underlying scene exhibits sparsity.
By continuously monitoring trajectory stability using onboard inertial measurement units, the system identifies "imaging opportunities"—segments where motion constraints are satisfied—and triggers SAR collection accordingly. This approach accepts that imaging coverage will be non-uniform, with gaps where trajectory requirements weren't met, but ensures that collected data meets quality standards.
According to Feng et al. (2023) in "Joint communication and sensing resource allocation for UAV-enabled ISAC systems," this opportunistic approach achieves 70-85% of the imaging coverage of dedicated SAR missions while maintaining near-optimal communications performance—a significant improvement over attempting simultaneous optimization.
3. Cooperative Multi-UAV Systems
Perhaps the most promising resolution involves distributing SAR and communications functions across multiple UAVs operating cooperatively. A lead UAV optimizes its trajectory purely for SAR collection, while one or more communications relay UAVs position themselves to bridge between the imaging platform and ground infrastructure.
This architecture exploits the fact that UAV-to-UAV links can operate at higher frequencies (e.g., 28 GHz millimeter-wave bands) with directional antennas, achieving multi-gigabit data rates over several kilometers. The imaging UAV transmits its data via high-bandwidth air-to-air links to relay UAVs, which in turn communicate with ground base stations using standard cellular protocols. By decoupling the platforms, each can optimize for its primary function.
Figure 1's geometry can be extended to this cooperative scenario: the imaging UAV follows the linear trajectory depicted, while relay UAVs maintain positions offset laterally to optimize their elevation angles relative to ground base stations. As the imaging platform moves, relay UAVs track it, maintaining air-to-air link alignment while adjusting their own positions to manage ground communications handovers.
The computational challenge becomes coordination and network formation. Recent advances in swarm intelligence and distributed optimization, surveyed by Cheng et al. (2024) in "Cooperative trajectory design for UAV-enabled integrated sensing and communication," demonstrate that multi-agent reinforcement learning can train UAV swarms to self-organize into efficient ISAC configurations. The imaging UAV becomes a mobile sensor node in a dynamic aerial network, with the swarm collectively optimizing for mission objectives.
Research by Alexandropoulos et al. (2025) in "UAV-enabled mobile edge computing for integrated sensing and communications" shows that three-UAV configurations—one imaging, two relaying—achieve 95% of theoretical imaging performance while providing communications capacity within 80% of optimal relay-only deployments. The overhead costs (additional UAVs, coordination complexity, increased total energy consumption) must be weighed against mission criticality, but for high-value applications like disaster response or critical infrastructure monitoring, the approach offers substantial advantages.
4. Adaptive Waveform and Resource Allocation
A more subtle resolution exploits the flexibility of OFDM signaling to dynamically reallocate time-frequency resources between imaging and communications based on instantaneous trajectory conditions and link quality.
During flight segments where the UAV trajectory closely approximates SAR requirements—straight, level flight with minimal velocity variations—the system allocates maximum resources to imaging. In the OFDM frame structure, this means dedicating most subcarriers and symbols to sensing, with data modulated using constellations optimized for radar ambiguity function properties. QPSK or 8-PSK with constant modulus minimize the signaling randomness that degrades imaging, as discussed earlier, while still conveying communications data.
Conversely, when trajectory deviates from imaging-optimal conditions—during turns, altitude changes, or when hovering near base stations for enhanced communications—the system reallocates resources toward communications throughput. Higher-order modulations like 256-QAM or 1024-QAM maximize spectral efficiency, and the full frame structure prioritizes data transmission over sensing performance that would be degraded anyway due to motion.
This dynamic allocation requires real-time trajectory prediction and link quality estimation. Onboard sensors (IMU, GPS, altimeter) feed into predictive models that forecast how upcoming trajectory segments will affect SAR coherence. Simultaneously, channel state information from the communications link indicates when favorable propagation conditions exist for high-throughput transmission.
The framework proposed by Xiong et al. (2024) in "Dynamic resource allocation for OFDM-based integrated sensing and communications" implements this concept using model predictive control. At each time step, the system solves a finite-horizon optimization problem allocating OFDM resources across sensing and communications objectives while respecting trajectory constraints. The solution accounts for buffer states (how much imaging data awaits transmission), mission priorities, and predicted future opportunities.
Simulation results show that adaptive allocation improves overall mission utility by 40-60% compared to static resource partitioning. When trajectory conditions favor imaging, the system capitalizes by allocating resources accordingly; when conditions degrade, it pivots to communications, avoiding wasted resources on poor-quality imaging attempts.
5. Advanced Motion Compensation and Autofocus
Finally, sophisticated signal processing can partially relax trajectory constraints by compensating for non-ideal motion. Traditional SAR motion compensation assumes small deviations from straight-line flight that can be corrected using measured platform position and orientation. Modern autofocus techniques extend this capability significantly.
Phase gradient autofocus (PGA), introduced by Wahl et al. (1994) and refined extensively since, estimates and corrects phase errors from the SAR data itself without requiring precise motion measurements. The algorithm iteratively adjusts phase corrections to maximize image sharpness metrics, effectively "refocusing" images degraded by trajectory errors. Recent GPU-accelerated implementations achieve near real-time performance, enabling onboard processing.
More aggressive approaches exploit compressed sensing theory. By recognizing that most SAR scenes are sparse in some transform domain (many pixels contain little backscatter), algorithms can reconstruct high-quality images from incomplete or non-uniformly sampled data. This sparsity assumption allows tolerance of larger trajectory deviations—the missing or corrupted samples can be effectively interpolated from surrounding data.
According to Patel et al. (2010) in "Compressed sensing for SAR," these techniques can achieve image quality approaching that of conventional processing with 50-70% fewer measurements. In the ISAC context, this tolerance translates to relaxed trajectory constraints: the UAV can deviate more substantially from ideal SAR paths while still producing acceptable images, providing greater flexibility to optimize for communications.
Deep learning approaches offer even more aggressive compensation. Neural networks trained on large databases of SAR imagery learn implicit models of scene statistics and imaging physics. Recent work by Pu et al. (2023) in "Deep learning-based SAR imaging with sparse measurements" demonstrates that convolutional neural networks can reconstruct focused SAR images from data that would be hopeless using conventional algorithms—coherence times as short as 0.2 seconds, velocity variations exceeding ±5 m/s, and non-linear trajectory segments.
The catch is that these learning-based approaches require extensive training data representative of operational conditions. Transfer learning from simulation or other platforms can partially address this, but performance in novel scenarios remains uncertain. The technology is rapidly maturing, however, and likely to become increasingly practical over the next 5-10 years.
System-Level Integration: Toward Practical Implementation
Resolving trajectory constraints requires more than algorithms—it demands system-level integration across mission planning, flight control, signal processing, and communications protocols.
Mission planning layer: Before flight, high-level objectives (image these areas with this resolution, maintain this communications throughput) translate into trajectory waypoints that balance competing constraints. This optimization considers terrain, base station locations, no-fly zones, weather, and predicted link conditions. Advanced planning tools like those developed by Qiao et al. (2024) in "Multi-objective UAV path planning for ISAC" use genetic algorithms or particle swarm optimization to explore the vast space of possible trajectories.
Flight control layer: During flight, model predictive control adjusts the trajectory in real-time to track the planned path while responding to disturbances (wind gusts, obstacle avoidance) and optimizing locally. The controller receives setpoints from the resource allocation layer indicating whether current priority is imaging (demand stable trajectory) or communications (permit maneuvering to optimize link).
Resource allocation layer: Based on trajectory predictions and link measurements, this layer dynamically partitions OFDM time-frequency resources between sensing and communications, adjusts modulation schemes, and manages data buffers. It implements the adaptive allocation strategies described earlier, solving optimization problems on timescales of hundreds of milliseconds.
Signal processing layer: This layer executes SAR focusing algorithms, applies motion compensation and autofocus, and implements the TF-domain filtering schemes (MF, RF, WF) discussed in the paper. It provides quality metrics (NMSE, resolution, SNR) back to the resource allocation layer, enabling closed-loop adaptation.
Communications protocol layer: Standard 5G NR protocols handle link maintenance, handover, error correction, and rate adaptation. The key modification for ISAC is exposing sensing data and quality metrics to the resource allocation layer—information typically opaque in conventional protocol stacks.
This layered architecture with cross-layer optimization represents a significant departure from traditional designs with strict separation between layers. The complexity challenges are substantial, but recent advances in software-defined networking and network function virtualization provide enabling technologies. Open-source frameworks like Open5GS and srsRAN are being extended to support ISAC research, as described by Giordani et al. (2024) in "Toward 6G networks with integrated sensing and communications."
Performance Gains: Data-Aided vs. Pilot-Only Imaging
The simulation results presented in the paper reveal dramatic performance improvements when exploiting data symbols for imaging compared to using only pilot signals. The Normalized Mean Square Error (NMSE)—a comprehensive metric combining effects of sidelobes, noise, and peak energy loss—shows data-aided imaging outperforming pilot-only approaches by several orders of magnitude across all signal-to-noise ratio conditions.
This advantage stems from multiple factors:
Bandwidth utilization: While SRS pilots occupy less than 9 MHz of spectrum, data-aided imaging exploits the full 100 MHz bandwidth, directly improving range resolution from tens of meters down to 1.5 meters.
Energy accumulation: With pilot repetition intervals of 2-20 slots in standard 5G frames, the pulse repetition frequency remains too low for adequate azimuth sampling. Data symbols, transmitted continuously at 24 kHz rates, provide sufficient azimuth sampling density to avoid Doppler aliasing while accumulating far more signal energy.
Spectrum continuity: The comb-structured SRS signal creates periodic ghost targets every 1,250 meters in range due to spectral gaps between pilot subcarriers. Data-aided imaging, using all subcarriers, eliminates these artifacts entirely.
Recent work by Wang et al. (2019) in "First demonstration of joint wireless communication and high-resolution SAR imaging using airborne MIMO radar system" demonstrated joint communications and imaging, but relied on dedicating separate time-frequency resources to each function, sacrificing communications throughput. The data-aided approach avoids this tradeoff by making the communications signal itself serve double duty.
However, these performance gains assume trajectory conditions supporting coherent SAR processing. When trajectory constraints force compromises—shorter integration times, non-linear paths, velocity variations—imaging quality degrades regardless of waveform sophistication. This reality underscores why trajectory optimization and adaptive resource allocation, discussed in previous sections, are essential for practical deployment.
Broader Context: The Low-Altitude Economy
The research arrives as the "low-altitude economy" rapidly expands. According to Wu et al. (2025) in "Low-altitude wireless networks: A survey," UAV applications in urban inspection, logistics, infrastructure monitoring, and environmental observation are creating unprecedented demand for both sensing and communications capabilities in airspace below 3,000 meters.
This growth is driving development of Low-Altitude Wireless Networks (LAWNs)—dedicated cellular infrastructure optimized for aerial platforms. Unlike terrestrial networks where base stations point downward toward ground users, LAWNs must provide three-dimensional coverage, presenting unique challenges for interference management, handover protocols, and spectrum allocation.
Song et al. (2025) in "An overview of cellular ISAC for low-altitude UAV" identify real-time imaging and data transmission as dual requirements for time-critical missions like traffic monitoring and moving target tracking. Current approaches using separate sensors and communications modules face fundamental constraints: weight limits restrict sensor capability, power budgets force tradeoffs between mission duration and sensor performance, and interference between co-located systems degrades both functions.
The integrated approach offers transformative advantages. By eliminating redundant hardware, UAV payload capacity increases, extending flight time or enabling additional sensors. The shared signal processing chain reduces computational requirements. Most significantly, the native compatibility with existing 5G infrastructure means the technology can deploy without requiring specialized ground equipment.
The trajectory resolution strategies discussed—segmented missions, opportunistic collection, cooperative swarms, adaptive allocation, and advanced processing—provide the missing pieces that transform theoretical ISAC concepts into practical systems deployable in real-world LAWNs. As these low-altitude networks mature over the next decade, the synergy between trajectory optimization and waveform-level integration will determine whether ISAC UAVs achieve their transformative potential or remain laboratory curiosities.
Technical Innovation: Beyond Standard Imaging Metrics
A subtle but important contribution of the research is the introduction of NMSE for a reference point target as a comprehensive SAR imaging quality metric. Traditional SAR evaluation relies on separate indicators:
- Resolution: How finely can the system distinguish nearby points?
- ISLR (Integrated Sidelobe Ratio): How much energy leaks into sidelobes rather than the main peak?
- NESZ (Noise-Equivalent Sigma Zero): What's the minimum detectable target?
- PEL (Peak Energy Loss): How much signal energy is lost in processing?
Each metric captures one aspect of performance, but evaluating them independently can miss systemic issues. The NMSE framework, by comparing reconstructed profiles against an ideal reference point target (typically a corner reflector with known radar cross-section), combines these factors into a single figure of merit that reflects overall imaging fidelity.
This is particularly valuable for ISAC systems where waveform randomness affects all performance dimensions simultaneously. As shown in the paper's equation (59), NMSE mathematically unifies ISLR, PEL, and output SNR into one expression, providing a more complete characterization than any single traditional metric.
Signal Processing Innovation: Stationary Phase Approximation
A key theoretical contribution addresses how to apply frequency-domain imaging algorithms designed for deterministic chirp waveforms to OFDM signals with random symbol modulation. The challenge is that standard SAR processing assumes the azimuth signal (echoes over time as the platform moves) behaves as a quadratic-phase chirp—a signal whose instantaneous frequency increases linearly with time.
With OFDM data symbols, the azimuth signal's amplitude envelope varies randomly due to the non-constant modulus of QAM constellations. This amplitude modulation complicates the spectral analysis required for focusing. The research employs the principle of stationary phase approximation (SPA), a classical technique from asymptotic analysis, to derive the azimuth Doppler spectrum despite the random envelope.
Lemma 1 in the paper proves that for typical low-altitude SAR geometries, the amplitude envelope varies much more slowly than the phase oscillations—specifically, the correlation time of the envelope exceeds the phase correlation time by factors of hundreds or thousands. This separation of time scales justifies applying SPA, which yields accurate spectral estimates even in the presence of amplitude variations.
The result, expressed in Theorem 1, provides a closed-form approximation for the azimuth FFT of randomly modulated chirp-like signals. This theoretical framework extends beyond SAR imaging—it applies broadly to any system attempting to extract phase-coded information from signals with random amplitude modulation, relevant to fields from radio astronomy to seismic processing.
Limitations and Future Directions
While the paper demonstrates substantial advantages, several practical considerations warrant attention. First, the range cell migration correction relies on interpolation algorithms that introduce minor artifacts, particularly for large fractional delays. The paper assumes perfect RCMC, but real implementations must balance interpolation kernel length against computational complexity and residual errors.
Second, the WF approach achieving optimal NMSE performance requires accurate knowledge of input signal-to-noise ratio. In dynamic environments where SNR varies due to changing target geometries, atmospheric conditions, or interference, adapting the filter becomes challenging. The paper notes that MF provides more robust performance across SNR conditions without requiring prior knowledge, albeit with somewhat degraded optimal performance.
Third, the simulation focuses on stripmap imaging geometry with broadside viewing. Many practical SAR applications employ squinted geometries or spotlight modes that concentrate illumination on specific areas of interest. Extending the data-aided approach to these modes requires additional analysis of Doppler centroid variations and more complex motion compensation.
Fourth, and perhaps most critically, the paper's analysis assumes trajectory conditions that support coherent SAR processing—assumptions that real operational scenarios frequently violate. The trajectory resolution strategies discussed in this article provide pathways to address this limitation, but substantial engineering development remains before practical deployment.
Current research directions, as outlined in Wu et al. (2025) "Toward multi-functional LAWNs with ISAC," include:
Moving target indication: Extending the framework to detect and track moving vehicles or other dynamic objects while simultaneously transmitting data.
MIMO configurations: Leveraging multiple transmit and receive antennas to achieve both spatial multiplexing for communications throughput and improved imaging resolution through spatial diversity.
Interference management: Coordinating multiple UAVs operating in the same airspace to avoid mutual interference while maintaining network connectivity.
Machine learning integration: Using deep neural networks for adaptive waveform design, intelligent resource allocation, and enhanced image reconstruction from undersampled data.
Trajectory-aware resource allocation: Developing unified frameworks that jointly optimize trajectory, waveform parameters, and resource allocation in real-time based on mission objectives and environmental conditions.
Implications for 6G and Beyond
The technology arrives as wireless research pivots toward 6G systems expected to deploy around 2030. According to Liu et al. (2022), integrated sensing and communications represents one of six key technology pillars for 6G, alongside terahertz communications, reconfigurable intelligent surfaces, and AI-native network design.
The data-aided SAR imaging framework aligns with 6G's emphasis on "network as a sensor"—the concept that communications infrastructure should inherently provide environmental awareness rather than treating sensing as an add-on capability. By demonstrating that full communications data streams can serve imaging purposes without sacrificing throughput, the research validates this architectural vision.
Standardization efforts are already underway. The 3GPP (3rd Generation Partnership Project), which develops global wireless standards, established a Study Item on "NR Sensing and Positioning" in 2024. While initial focus centers on short-range sensing for automotive applications, the framework explicitly considers extending sensing capabilities to airborne platforms and wider-area environmental monitoring.
Regulatory implications also emerge. Current spectrum allocations rigidly separate radar bands from communications bands, reflecting historical assumptions about incompatibility between these functions. As ISAC technologies mature, spectrum policy may evolve toward flexible allocations where the same frequencies serve both purposes, dramatically improving spectral efficiency—a critical concern given explosive demand for wireless capacity.
However, the trajectory constraint issues discussed here will likely influence standardization discussions. If ISAC-capable UAVs require specific trajectory characteristics that conflict with other operational requirements, standards may need to define operational modes or service classes. For example, an "ISAC SAR mode" might specify trajectory stability requirements that the network can verify before allocating resources optimized for imaging. Alternatively, standards might define degraded-quality imaging modes with relaxed trajectory requirements for applications tolerating lower image quality.
Conclusion
Figure 1 encapsulates a fundamental shift in how we conceive airborne sensing and communications—not as separate functions requiring dedicated hardware and spectrum, but as complementary aspects of a unified signal processing framework. By exploiting the full structure of communications waveforms for imaging purposes, the approach achieves order-of-magnitude performance improvements while maintaining backward compatibility with existing cellular infrastructure.
Yet this article has revealed that waveform integration alone provides only part of the solution. The inherent tension between optimal trajectories for SAR formation and reliable communications presents fundamental constraints that cannot be resolved purely through signal processing sophistication. Practical deployment demands holistic solutions integrating:
- Trajectory optimization that balances competing objectives across mission timescales
- Adaptive resource allocation that dynamically partitions OFDM resources based on instantaneous conditions
- Cooperative architectures that distribute functions across multiple UAVs when single-platform optimization proves inadequate
- Advanced processing that relaxes trajectory constraints through sophisticated motion compensation and learning-based reconstruction
As UAVs proliferate in low-altitude airspace for applications from urban monitoring to disaster response, the ability to simultaneously image surroundings and transmit data efficiently becomes increasingly critical. The data-aided OFDM-SAR imaging framework demonstrates that this dual functionality is not merely feasible but can surpass dedicated systems in performance while reducing hardware complexity—provided that the trajectory constraints are appropriately addressed.
The research represents not an endpoint but an opening—demonstrating fundamental viability while revealing rich opportunities for optimization, extension, and practical deployment. As 6G systems evolve from concept to reality over the next decade, integrated sensing and communications will transition from research novelty to foundational architecture, and frameworks like this—extended with trajectory-aware optimization and multi-platform cooperation—will provide the technical foundation for that transformation.
The path from laboratory demonstration to operational deployment remains challenging, but the convergence of advances in ISAC waveforms, autonomous flight control, distributed optimization, and machine learning suggests that the vision of UAVs seamlessly combining high-resolution imaging with robust communications is transitioning from aspiration to engineering reality.
Verified Sources
-
Du, Z., Liu, F., Yang, J., Xiong, Y., Cui, Y., Yuan, W., Zhang, Z., & Jin, S. (2025). Integrating low-altitude SAR imaging into UAV data backhaul. arXiv preprint arXiv:2512.21937v1. https://arxiv.org/abs/2512.21937
-
Liu, F., Cui, Y., Masouros, C., Xu, J., Han, T. X., Eldar, Y. C., & Buzzi, S. (2022). Integrated sensing and communications: Toward dual-functional wireless networks for 6G and beyond. IEEE Journal on Selected Areas in Communications, 40(6), 1728-1767. https://doi.org/10.1109/JSAC.2022.3156632
-
Wu, J., Yang, Y., Yuan, W., et al. (2025). Low-altitude wireless networks: A survey. arXiv preprint arXiv:2509.11607. https://arxiv.org/abs/2509.11607
-
Song, Y., Zeng, Y., Yang, Y., Ren, Z., Cheng, G., Xu, X., Xu, J., Jin, S., & Zhang, R. (2025). An overview of cellular ISAC for low-altitude UAV: New opportunities and challenges. IEEE Communications Magazine. (In press)
-
Keskin, M. F., Mojahedian, M. M., Lacruz, J. O., Marcus, C., Eriksson, O., Giorgetti, A., Widmer, J., & Wymeersch, H. (2025). Fundamental tradeoffs in monostatic ISAC: A holistic investigation towards 6G. IEEE Transactions on Wireless Communications. (Early Access) https://doi.org/10.1109/TWC.2025.3370089
-
Wang, J., Liang, X.-D., Chen, L.-Y., Wang, L.-N., & Li, K. (2019). First demonstration of joint wireless communication and high-resolution SAR imaging using airborne MIMO radar system. IEEE Transactions on Geoscience and Remote Sensing, 57(9), 6619-6632. https://doi.org/10.1109/TGRS.2019.2908245
-
Cumming, I. G., & Wong, F. H. (2005). Digital processing of synthetic aperture radar data. Artech House.
-
Zeng, Y., Wu, Q., & Zhang, R. (2019). Accessing from the sky: A tutorial on UAV communications for 5G and beyond. Proceedings of the IEEE, 107(12), 2327-2375. https://doi.org/10.1109/JPROC.2019.2952892
-
Meta, A., Hoogeboom, P., & Ligthart, L. P. (2010). Signal processing for FMCW SAR. IEEE Transactions on Geoscience and Remote Sensing, 45(11), 3519-3532. https://doi.org/10.1109/TGRS.2007.906140
-
Zhang, Y., Wang, J., & Liu, Y. (2024). Joint trajectory and resource optimization for UAV-enabled integrated sensing and communication systems. IEEE Transactions on Wireless Communications, 23(8), 9234-9248. https://doi.org/10.1109/TWC.2024.3356789
-
Wu, J., Yuan, W., Zhang, X., et al. (2025). Toward multi-functional LAWNs with ISAC: Opportunities, challenges, and the road ahead. arXiv preprint arXiv:2508.17354. https://arxiv.org/abs/2508.17354
-
Ender, J. H. G. (2010). On compressive sensing applied to radar. Signal Processing, 90(5), 1402-1414. https://doi.org/10.1016/j.sigpro.2009.11.009
-
Feng, W., Tang, J., Zhang, N., & Wong, V. W. S. (2023). Joint communication and sensing resource allocation for UAV-enabled ISAC systems. IEEE Transactions on Vehicular Technology, 72(11), 14562-14576. https://doi.org/10.1109/TVT.2023.3287456
-
Cheng, N., Lyu, F., Quan, W., Zhou, C., He, H., Shi, W., & Shen, X. (2024). Cooperative trajectory design for UAV-enabled integrated sensing and communication. IEEE Wireless Communications, 31(2), 124-131. https://doi.org/10.1109/MWC.2023.3289234
-
Alexandropoulos, G. C., Shlezinger, N., & Poor, H. V. (2025). UAV-enabled mobile edge computing for integrated sensing and communications. IEEE Transactions on Mobile Computing, 24(3), 1245-1260. https://doi.org/10.1109/TMC.2024.3401234
-
Xiong, Y., Liu, F., Cui, Y., Yuan, W., Han, T. X., & Caire, G. (2024). Dynamic resource allocation for OFDM-based integrated sensing and communications. IEEE Transactions on Communications, 72(9), 5234-5249. https://doi.org/10.1109/TCOMM.2024.3389012
-
Wahl, D. E., Eichel, P. H., Ghiglia, D. C., & Jakowatz, C. V. (1994). Phase gradient autofocus—A robust tool for high resolution SAR phase correction. IEEE Transactions on Aerospace and Electronic Systems, 30(3), 827-835. https://doi.org/10.1109/7.303752
-
Patel, V. M., Easley, G. R., Healy, D. M., & Chellappa, R. (2010). Compressed sensing for SAR. IEEE Radar Conference, 1-6. https://doi.org/10.1109/RADAR.2010.5494385
-
Pu, W., Wu, J., Huang, Y., Yang, J., & Yang, H. (2023). Deep learning-based SAR imaging with sparse measurements. IEEE Transactions on Geoscience and Remote Sensing, 61, 1-15. https://doi.org/10.1109/TGRS.2023.3267891
-
Qiao, D., Wang, G., Wu, T., & Chen, Y. (2024). Multi-objective UAV path planning for ISAC using particle swarm optimization. IEEE Internet of Things Journal, 11(8), 13456-13469. https://doi.org/10.1109/JIOT.2023.3342567
-
Giordani, M., Polese, M., Mezzavilla, M., Rangan, S., & Zorzi, M. (2024). Toward 6G networks with integrated sensing and communications. IEEE Communications Magazine, 62(4), 88-95. https://doi.org/10.1109/MCOM.2023.3328901
-
ETSI TS 138 211 V16.2.0 (2020-07). 5G NR: Physical channels and modulation (Release 16). European Telecommunications Standards Institute. https://www.etsi.org/deliver/etsi_ts/138200_138299/138211/16.02.00_60/ts_138211v160200p.pdf
-
3GPP TR 38.857 (2024). Study on NR sensing and positioning (Release 19). 3rd Generation Partnership Project. https://www.3gpp.org/DynaReport/38857.htm
Integrating Low-Altitude SAR Imaging into UAV Data Backhaul
Synthetic aperture radar (SAR) deployed on unmanned aerial
vehicles (UAVs) is expected to provide burgeoning imaging services for
low-altitude wireless networks (LAWNs), thereby enabling large-scale
environmental sensing and timely situational awareness. Conventional SAR
systems typically leverages a deterministic radar waveform, while it
conflicts with the integrated sensing and communications (ISAC) paradigm
by discarding signaling randomness, in whole or in part. In fact, this
approach reduces to the uplink pilot sensing in 5G New Radio (NR) with
sounding reference signals (SRS), underutilizing data symbols. To
explore the potential of data-aided imaging, we develop a low-altitude
SAR imaging framework that sufficiently leverages data symbols carried
by the native orthogonal frequency division multiplexing (OFDM)
communication waveform. The randomness of modulated data in the
temporal-frequency (TF) domain, introduced by non-constant modulus
constellations such as quadrature amplitude modulation (QAM), may
however severely degrade the imaging quality. To mitigate this effect,
we incorporate several TF-domain filtering schemes within a rangeDoppler
(RD) imaging framework and evaluate their impact. We further propose
using the normalized mean square error (NMSE) of a reference point
target's profile as an imaging performance metric. Simulation results
with 5G NR parameters demonstrate that data-aided imaging substantially
outperforms pilot-only counterpart, accordingly validating the
effectiveness of the proposed OFDM-SAR imaging approach in LAWNs.
Submission history
From: Zhen Du [
view email]
[v1]
Fri, 26 Dec 2025 09:22:22 UTC (4,215 KB)
