Formation Control for CRLB-Optimal Cooperative Sensing in Low-Altitude Wireless Networks

arXiv:2602.23990 · eess.SP · Submitted 2026-02-27 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Electrical Engineering. The day in electrical engineering: signals, systems and the hardware that carries them.

Rosa: We're looking at the paper: "Formation Control for CRLB-Optimal Cooperative Sensing in Low-Altitude Wireless Networks".

Dev: Here's the gist of the paper. Cooperative sensing with uncrewed aerial vehicles (UAVs) is a key enabler for low-altitude wireless networks (LAWNs), where sensing accuracy critically depends on the spatial configuration of the UAV formation.

Rosa: The paper studies formation design and control for Cramér–Rao lower bound (CRLB)-optimal cooperative target sensing.

Dev: Sensing Performance Modeling The study first establishes a sensing performance model based on range measurements and derives the Fisher information matrix (FIM) of the target location.

Rosa: The variance of the measurement noise is modeled as being inversely proportional to the signal-to-noise ratio (SNR) at the UAV.

Title and authors: Rosa: So, this paper is about formation control for CRLB-optimal cooperative sensing in low-altitude wireless networks. It really focuses on how you should arrange the UAVs to get the best possible sensing accuracy when they’re flying low.

Dev: Yeah, exactly. They are looking at formation design and control specifically for Cramér–Rao lower bound optimal cooperative target sensing, which just means they're trying to figure out the best way to place those drones so you can estimate where a ground target is located with the least possible error variance.

Taro: I wonder if this setup actually works outside of a perfect lab scenario; how robust is this formation when things get messy in the real world?

Rosa: That's a fair question, Taro. The paper models sensing performance based on range measurements and derives what they call the Fisher information matrix or FIM for the target location. It also models measurement noise variance as being inversely proportional to the signal-to-noise ratio at the UAV, with a specific formula involving kappa, sigma zero squared, and dm four to power m plus beta zero.

Dev: That noise model tells us how noisy our measurements are based on how strong the signal is where the UAV is located. And then they figure out that for a two-way sensing channel, the power gain follows a quartic distance loss.

Rosa: Right, so they set up a performance model first before diving into the actual formation design. Then, when it comes to finding the best geometry, they use an A-optimality criterion to minimize that average estimation variance. They find that for this minimization to work out perfectly, the Fisher information matrix needs to be isotropic in the horizontal plane.

Dev: That requirement leads them directly to a specific geometric characterization of the optimal sensing formation F star, which turns out to be an inverted cone with its vertex sitting right at where the target is located.

Taro: An inverted cone sounds spatially intuitive, but I’m curious about how much flexibility this fixed shape gives us when the target isn't perfectly known or when we need to adapt quickly.

Rosa: The geometry is defined by elevation angles phi m and azimuth angles theta m. Proposition one shows that the trace of the inverse Fisher information matrix reaches its global minimum if and only if the FIM becomes a scalar matrix, which happens when the eigenvalues are equal, leading to a regular polygon geometry with a common elevation angle.

Title and authors: Dev: But they don't stop there; they then develop this distributed formation control strategy to steer those UAVs from any starting point towards that sensing-optimal configuration. This control input is designed as a distributed negative gradient step of the displacement error term in the cost function.

Taro: A distributed gradient step sounds manageable, but what about real-time execution? If there's latency or if one UAV loses its link, how does this control strategy handle those failures gracefully?

Rosa: They designed it to leverage linear closed-loop dynamics so that global convergence is guaranteed even if you only have local neighbor information. Plus, they integrated a repulsive control component to actively prevent obstacle collisions in practice.

Dev: That integration of the repulsive term is important for the safety case; it means the system isn't just optimizing for sensing accuracy in a vacuum; it's managing physical constraints while achieving that optimal formation shape.

Taro: So, when the world misbehaves, like an obstacle suddenly appears, how does this distributed control react? Does it prioritize maintaining the desired sensing configuration or avoiding a crash first?

Rosa: The repulsive component is specifically there to handle those collisions by pushing the drones away from obstacles. And because they use linear closed-loop dynamics, it should manage that trade-off reasonably well during formation reshaping.

Dev: The numerical validation shows that this proposed CRLB optimal formation F star consistently outperforms benchmark schemes across the entire altitude range studied, and that advantage gets bigger as altitude increases.

Taro: Why does it get better at higher altitudes? Is it because the path loss effects amplify, or is there some other factor related to the geometry that helps?

Rosa: The paper suggests that this performance increase is due to the fact that higher altitudes amplify path-loss effects and reduce angular diversity in a way that benefits this specific sensing setup. It’s a trade-off between those two factors.

Dev: And they showed how their distributed control strategy effectively couples the reshaping of the formation with the actual cooperative motion, which is key for keeping everything synchronized during this process.

Taro: That coupling sounds complex to implement reliably in real time, Rosa. How do you handle the dynamic adjustment of that target displacement set F? Does it just scale up and down, or is there a specific mechanism?

Title and authors: Rosa: They propose an improvement where the system can dynamically adjust the target displacement set F by incorporating both a global scaling factor and a repulsive component. This allows for navigation through narrow constraints by reducing that scaling factor when they get close to spatial bottlenecks.

Dev: That dynamic adjustment sounds like it adds a layer of robustness against tight maneuvering, which is something I’d want to see tested in real-time loops. But we have to be careful about the control loop rate when dealing with these changing factors.

Taro: So, the paper shows that this formation control for CRLB optimal cooperative sensing in low altitude wireless networks uses a specific geometric characterization, a distributed gradient step for control, and a dynamic adjustment mechanism for navigation.

Rosa: That's right; it provides a solid framework where you optimize the shape based on minimizing the error variance, then you steer to that shape using local information while actively avoiding crashes.

Dev: Overall, this work gives us a concrete method for designing UAV formations not just based on simple coverage, but specifically tailored to maximize sensing accuracy under noisy conditions in low-altitude environments.

Taro: For someone who only listens to the show, what does this mean practically? It means that if you're running a low-altitude network for sensing, you don't have to guess the best shape; this method tells you exactly what the shape should be based on your path loss and noise parameters.

Rosa: Exactly. And it means that when things go wrong, the control system has a built-in way to steer back toward that optimal state while keeping those physical obstacles out of the way.

Dev: It’s about moving from just having some UAVs flying together to having a system actively shaping itself for peak performance in a noisy, constrained environment.

Taro: I just hope the simulation results translate well when you put this on actual hardware where you can see those path loss and diversity effects happening live.

Rosa: We'll have to see how it holds up when the world misbehaves in real-time trials, but the theoretical foundation here seems pretty solid for guiding that development.

Dev: That's what we need to keep an eye on as we look at future work stemming from this paper.

Taro: Alright, let's move on to another interesting piece of research.

The paper's summary: Rosa: So, we’re looking at how the authors summarized their paper on formation control for cooperative sensing in low altitude wireless networks and what that actually means for us out here on the ground.

Dev: Basically, they start by saying they modeled things using range measurements to figure out the Fisher information matrix, which is like a map of where a target might be, and then they derived the noise variance based on how strong the signal is at your UAV.

Rosa: And then they figured out that for getting the best estimate—the Cramér–Rao lower bound—the optimal formation has to be arranged so that its Fisher information matrix is isotropic in the horizontal plane.

Dev: Which means geometrically, the ideal shape turns into an inverted cone with its tip right at where you’re looking for the target. It’s a very specific geometric requirement they set up using what they call A-optimality criterion to minimize the average estimation variance.

Rosa: That setup gives them a really regular polygon geometry with all the UAVs sharing a common elevation angle, which is important because it balances that path loss effect against the angular diversity you get from having multiple angles.

Dev: Then they moved into the control part, where they designed a distributed formation control strategy that lets the drones move from wherever you start to this optimal shape. They use a distributed negative gradient step of the displacement error term in the cost function for that steering.

Rosa: And here’s where it gets practical for me: they made sure this control works even with only local neighbor information, using linear closed-loop dynamics, and they even added a repulsive control component to keep them away from obstacles.

Dev: That safety aspect is crucial because in real-time loops, you need that stability. The numerical validation showed this optimal formation consistently beats other schemes across all altitudes studied.

Rosa: But the results also hinted that the performance gap gets bigger as altitude increases, which they attribute to how path loss and angular diversity change up there.

Dev: So, what does this mean for us? It means we don't just guess a shape; this method gives us a way to calculate exactly what the optimal shape should be based on our specific noise levels and signal conditions.

Rosa: And when things get messy—like obstacles or moving targets—the control strategy is built to actively steer back toward that best sensing shape while keeping those physical constraints in mind.

Dev: It’s about moving from just having some drones flying together to a system that’s actively shaping itself for peak accuracy in a noisy, constrained environment.

Rosa: Next up, we're going to look at how this formation control strategy handles the dynamic adjustment of the target displacement set F when things get tight.

The paper's improvements: Tom: So, we’re talking about how the authors suggest they can make this sensing system even better by adding some improvements to their original design.

Rosa: They propose a few key things starting with optimizing that target displacement set F dynamically, which means it doesn't just stay fixed. Instead, it incorporates both a global scaling factor and that repulsive component we talked about earlier.

Dev: That dynamic adjustment is designed so the UAVs can navigate through really narrow constraints better; they reduce that scaling factor when they get close to those spatial bottlenecks.

Rosa: So it’s an active way to handle tight maneuvering, trying to keep the formation shape optimal even when space is limited. It sounds like something we could test on a field robot later.

Dev: From a controls standpoint, that dynamic adjustment adds complexity to the loop rate because you have to account for those changing factors in real-time. We'd need a very fast system to make sure that doesn't introduce latency issues where it matters most.

Rosa: That’s true, but the paper argues this makes the whole process more robust against unexpected physical limits, which is exactly what I want to see when we take these trials outside the lab.

Dev: And they also mentioned how this dynamic adjustment helps manage those trade-offs between path loss and angular diversity more intelligently than a static setup could. It’s about balancing those two factors in real time.

Rosa: So, the implication is that if you can implement that dynamic set F adjustment, the system should be able to maintain near-optimal sensing performance even when things get physically restrictive.

Dev: Exactly; it keeps the formation shape from getting stuck in a sub-optimal configuration just because of some immediate physical interference. It’s about graceful recovery in motion.

Rosa: It shows that for low altitude, the geometry isn't static; it needs to be adaptive to the environment around you.

Dev: And if we can nail that adaptation, then this framework could become much more versatile for various sensing tasks in dense urban or outdoor settings. What’s next is seeing how they actually tested these improvements against the original model parameters.

Conclusion: Tom: So we’re wrapping up this discussion on the paper "Formation Control for CRLB-Optimal Cooperative Sensing in Low-Altitude Wireless Networks" by summarizing what they actually proved and where it takes us next.

Rosa: Basically, they showed that by using that specific geometric shape—the inverted cone—and combining it with a distributed control strategy, you can achieve the lowest possible estimation error variance for sensing targets in low altitude networks.

Dev: That means we’ve got a clear blueprint for designing the UAV formation to maximize our data quality while keeping the drones safe from things like obstacles. The safety case hinges on that distributed control working reliably with only local neighbor information.

Taro: I wonder if this works well when the environment gets really chaotic, like sudden wind gusts or unexpected interference. Does that inverted cone shape hold up under those kinds of sudden shifts?

Rosa: The paper shows it performs well across a wide range of altitude, but they note that performance can drop as altitude increases because path loss and angular diversity both change in ways that hurt the system.

Dev: And if you look at the improvements they suggested, like adding that dynamic target displacement set F adjustment, it’s an attempt to make the system more resilient to those real-world fluctuations.

Taro: That sounds promising because it means the formation isn't just a fixed shape anymore; it can actively change its configuration based on what's happening in space right now.

Rosa: It shifts the design from a static optimal shape to a dynamic one that responds to spatial constraints, which is exactly what field trials need for real-world deployment.

Dev: So for anyone listening who cares about real-time loops, it suggests that incorporating those scaling factors into the control input is necessary if you want to maintain high accuracy when navigating complex spaces.

Taro: It’s a good direction for autonomy research because it moves the problem from just "find the best shape" to "keep this shape optimal while moving through obstacles."

Rosa: Exactly, and by looking at how they handled those dynamic adjustments, we get a better idea of what’s needed for future field robotic deployments.

Dev: Overall, this work provides a strong theoretical foundation for designing UAV swarms that are both accurate in their sensing and safe in their control actions. It’s solid engineering work.

Taro: I think the next step is testing those improvements under more extreme scenarios than what they studied, to see if that dynamic adjustment really keeps things stable when the world misbehaves.

Jun Wu, Haijia Jin, Nanchi Su, Jinna Li, Haoyuan Pan, Tse-Tin Chan

Southern University of Science and Technology · The Education University of Hong Kong

eess.SP

Submitted: 2026-02-27

Updated: 2026-02-27

DOI: 10.1109/TVT.2026.3743628

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 76/100

The gist: Cooperative sensing with uncrewed aerial vehicles (UAVs) is a key enabler for low-altitude wireless networks (LAWNs)…

Key concepts

Fisher Information Matrix (FIM)
The FIM quantifies the amount of information a set of measurements provides about an unknown parameter, such as the target's location. In this study, it is used to model sensing performance and determine the optimal formation geometry required to maximize this information.
CRLB-optimal Formation
This refers to designing the UAV positions such that the minimum possible estimation error (the CRLB) is achieved. The paper finds that an inverted cone shape, characterized by a common elevation angle, provides this optimality under specific mathematical conditions related to the Fisher information matrix.
Distributed Formation Control
This strategy manages multiple UAVs simultaneously. Instead of one central controller, each UAV uses information only from its immediate neighbors to adjust its movement. This allows the entire formation to move towards the optimal sensing configuration while coordinating motion and avoiding collisions.
A-optimality Criterion
This criterion is used to select the best formation geometry by minimizing the average estimation variance of the target location. It mathematically guides the determination of a regular polygon shape with a specific elevation angle that balances sensing accuracy against path loss effects.

Terminology

Summary

Cooperative sensing with uncrewed aerial vehicles (UAVs) is a key enabler for low-altitude wireless networks (LAWNs), where sensing accuracy critically depends on the spatial configuration of the UAV formation. p. 0

Where the paper says it

1 Formation Control for CRLB-Optimal Cooperative Sensing in Low-Altitude Wireless Networks Jun Wu, Student Member, IEEE, Haijia Jin, Student Member, IEEE, Nanchi Su, Member, IEEE, Jinna Li, Senior Member, IEEE, Haoyuan Pan, Member, IEEE, and Tse-Tin Chan, Member, IEEE Abstract—Cooperative sensing with uncrewed aerial vehicles (UAVs) is a key enabler for low-altitude wireless networks (LAWNs), where sensing accuracy critically depends on the spatial configuration of the UAV formation. In this paper, we study formation design and control for Cramér–Rao lower bound (CRLB)-optimal cooperative target sensing.

The paper studies formation design and control for Cramér–Rao lower bound (CRLB)-optimal cooperative target sensing. p. 0
Where the paper says it

1 Formation Control for CRLB-Optimal Cooperative Sensing in Low-Altitude Wireless Networks Jun Wu, Student Member, IEEE, Haijia Jin, Student Member, IEEE, Nanchi Su, Member, IEEE, Jinna Li, Senior Member, IEEE, Haoyuan Pan, Member, IEEE, and Tse-Tin Chan, Member, IEEE Abstract—Cooperative sensing with uncrewed aerial vehicles (UAVs) is a key enabler for low-altitude wireless networks (LAWNs), where sensing accuracy critically depends on the spatial configuration of the UAV formation. In this paper, we study formation design and control for Cramér–Rao lower bound (CRLB)-optimal cooperative target sensing.

Sensing Performance Modeling

The study first establishes a sensing performance model based on range measurements and derives the Fisher information matrix (FIM) of the target location. The variance of the measurement noise is modeled as being inversely proportional to the signal-to-noise ratio (SNR) at the UAV. p. 1

Where the paper says it

The distance vector is denoted by d = [d1, . In general, the variance σ 2 m is inversely proportional to the signal-to-noise ratio (SNR) at the UAV, i.

For a two-way sensing channel, the power gain is modeled as a quartic distance loss. p. 1
Where the paper says it

, SNRm = pmGphm σ 2 0 , where pm is the transmit power allocated to UAV m, Gp is the processing gain, and σ 2 0 is the receiver noise power. For a two-way sensing channel, the power gain is modeled as a quartic distance loss, i.

The measurement noise variance can be particularized as sigma 2m = κσ 2 0(dm) 4pmGpβ0, where κ being a constant system parameter. p. 1
Where the paper says it

Here, GT and Gr denote the transmit and receive antenna gains, respectively, σrcs is the radar cross section (RCS), and λ is the carrier wavelength. Consequently, σ 2 m can be particularized as σ 2 m = κσ2 0 (dm) 4 pmGpβ0 , with κ being a constant system parameter [7].

Optimal Formation Geometry Derivation

The optimal formation is analytically characterized by adopting the A-optimality criterion, which minimizes the average estimation variance. Proposition 1 suggests that tr(J−1s) achieves minimal value P 4M m=1 wm if Jxx = Jyy and Jxy = 0. p. 2

Where the paper says it

When λ1 = λ2 = λ, the diagonal matrix becomes Λ = λI, which yields Js = U(λI)UT = λI, completing the proof. Proposition 1 suggests that tr(J −1 s ) achieves minimal value P 4 M m=1 wm if Jxx = Jyy and Jxy = 0.

This leads to the determination of the optimal formation F∗ being reformulated as an optimization problem involving elevation angles ϕm. p. 2
Where the paper says it

Proposition 1 suggests that tr(J −1 s ) achieves minimal value P 4 M m=1 wm if Jxx = Jyy and Jxy = 0. In other words, the determination of the optimal formation F ∗ can be reformulated as the following optimization problem min ϕm,θm 4 PM m=1 wm (12a) s.

The optimal azimuth angles are determined by the condition that the Fisher information must be isotropically distributed in the horizontal plane to eliminate off-diagonal correlations in the FIM. p. 2
Where the paper says it

This condition implies that the Fisher information must be isotropically distributed in the horizontal plane to eliminate off-diagonal correlations in the FIM. A standard solution satisfying these equations is the regular M-polygon formation, where the azimuth angles are distributed as θ ∗ m = θ0 + 2π(m−1) M , m = 1, .

In summary, the optimal sensing formation F∗ can be geometrically characterized as an inverted cone with its vertex at the target’s location. p. 2
Where the paper says it

, M, where θ0 ∈ [0, 2π) is an arbitrary initial rotation angle. In summary, the optimal sensing formation F ∗ can be geometrically characterized as an inverted cone with its vertex at the target’s location.

Distributed Formation Control Strategy

A distributed formation control strategy is developed to steer UAVs from arbitrary initial deployments toward the desired formation while ensuring coordinated motion and obstacle avoidance. The control input um,k is designed as a distributed negative-gradient step of the displacement-error term in the cost function. p. 3

Where the paper says it

Hence, the total cost function is given by Ltot = P m∈M Lm (F, qm,k). Motivated by (15), we design the formation-shape control input um,k as a distributed negative-gradient step of the displacement-error term in (15) with respect to qm,k, which yields the control policy um,k = −ε X p∈M amp (qm,k − qp,k − δpm), (16) with ε > 0 being the convergence speed tuning parameter.

This approach leverages linear closed-loop dynamics to guarantee global convergence where only local neighbor information is required. p. 3
Where the paper says it

Motivated by (15), we design the formation-shape control input um,k as a distributed negative-gradient step of the displacement-error term in (15) with respect to qm,k, which yields the control policy um,k = −ε X p∈M amp (qm,k − qp,k − δpm), (16) with ε > 0 being the convergence speed tuning parameter. This displacement-based approach leverages linear closed-loop dynamics to guarantee global convergence, where only local neighbor information is required, resulting in low computational complexity and communication overhead.

To prevent obstacle collisions in practice, a repulsive control component ur m,k is further integrated into the control architecture. p. 3
Where the paper says it

This displacement-based approach leverages linear closed-loop dynamics to guarantee global convergence, where only local neighbor information is required, resulting in low computational complexity and communication overhead. To prevent obstacle collisions in practice, a repulsive control component u r m,k is further integrated into the control architecture [12].

Numerical Validation and Results

Extensive simulation results verify the effectiveness of the proposed algorithm. The proposed CRLB-optimal formation F∗ consistently outperforms benchmark schemes across the entire altitude range. p. 3

Where the paper says it

2 depicts the CRLB as a function of the UAV flight altitude under different formation configurations, thereby validating the effectiveness of the proposed cooperative sensing–oriented formation design. It is observed that the proposed CRLB-optimal formation F ∗ consistently outperforms the benchmark schemes across the entire altitude range.

The performance advantage becomes more pronounced as the altitude increases, which can be attributed to the fact that higher altitudes amplify path-loss effects and reduce angular diversity. p. 3
Where the paper says it

It is observed that the proposed CRLB-optimal formation F ∗ consistently outperforms the benchmark schemes across the entire altitude range. Moreover, the performance advantage becomes more pronounced as the altitude increases, which can be attributed to the fact that higher altitudes amplify path-loss effects and reduce angular diversity, making the sensing accuracy increasingly sensitive to the geometric configuration of the formation.

The evolution process demonstrates that the proposed distributed control strategy effectively couples formation reshaping with cooperative motion. p. 4
Where the paper says it

Once the obstacles are safely bypassed, the formation progressively expands and reconfigures to recover the prescribed sensing-optimal geometry F ∗ and eventually hovers over the target. This evolution process demonstrates that the proposed distributed control strategy effectively couples formation reshaping with cooperative motion.

Conclusion

This paper studied CRLB-optimal formation design and distributed control for cooperative sensing in LAWNs. By analyzing range-based measurements, the Fisher information matrix for the target location was derived. p. 4

Where the paper says it

CONCLUSION This paper studied CRLB-optimal formation design and distributed control for cooperative sensing in LAWNs. By analyzing range-based measurements, the Fisher information matrix for the target location was derived, and the sensingoptimal formation was analytically characterized under the A-optimality criterion, yielding a regular polygon geometry with a common elevation angle.

The sensingoptimal formation was analytically characterized under the A-optimality criterion, yielding a regular polygon geometry with a common elevation angle. p. 4
Where the paper says it

CONCLUSION This paper studied CRLB-optimal formation design and distributed control for cooperative sensing in LAWNs. By analyzing range-based measurements, the Fisher information matrix for the target location was derived, and the sensingoptimal formation was analytically characterized under the A-optimality criterion, yielding a regular polygon geometry with a common elevation angle.

Then, a distributed formation control strategy was developed to navigate UAVs from arbitrary initial positions toward the sensing-optimal configuration while ensuring obstacle avoidance. p. 4
Where the paper says it

By analyzing range-based measurements, the Fisher information matrix for the target location was derived, and the sensingoptimal formation was analytically characterized under the A-optimality criterion, yielding a regular polygon geometry with a common elevation angle. Then, a distributed formation control strategy was developed to navigate UAVs from arbitrary initial positions toward the sensing-optimal configuration while ensuring obstacle avoidance.

Extensive simulation results verified the effectiveness of the proposed design. p. 4
Where the paper says it

Then, a distributed formation control strategy was developed to navigate UAVs from arbitrary initial positions toward the sensing-optimal configuration while ensuring obstacle avoidance. Extensive simulation results verified the effectiveness of the proposed design.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. The AI system can be designed to optimize UAV formation geometry for sensing tasks by minimizing Cramér–Rao lower bound (CRLB) based on an isotropic Fisher information matrix, leading to a regular polygon geometry with an elevation angle determined by the tradeoff between path loss and geometric diversity.

  2. The AI system can implement a distributed formation control strategy that steers UAVs from arbitrary initial deployments toward the sensing-optimal configuration while maintaining formation motion and obstacle avoidance using only local neighbor information.

  3. The system can dynamically adjust the target displacement set F, incorporating both a global scaling factor and a repulsive component, allowing for formation navigation through narrow constraints by reducing the scaling factor when approaching spatial bottlenecks.

Abstract

Cooperative sensing with uncrewed aerial vehicles (UAVs) is a key enabler for low-altitude wireless networks (LAWNs), where sensing accuracy critically depends on the spatial configuration of the UAV formation. In this paper, we study formation design and control for Cramer-Rao lower bound (CRLB)-optimal cooperative target sensing. We first establish a sensing performance model based on range measurements and derive the Fisher information matrix (FIM) of the target location. By adopting the A-optimality criterion, we analytically characterize the formation geometry that minimizes the CRLB of the estimation error. The optimal formation is shown to exhibit isotropic Fisher information in the horizontal plane, leading to a regular polygon geometry with an elevation angle determined by the tradeoff between path loss and geometric diversity. Building on this result, we further develop a distributed formation control strategy that steers UAVs from arbitrary initial deployments toward the sensing-optimal configuration while maintaining formation motion and obstacle avoidance. Numerical results demonstrate that the proposed scheme consistently outperforms benchmark formations in terms of CRLB and achieves reliable convergence under practical constraints.

Sources

Related papers