⚡ electrical engineering

DOA Estimation for Low-Altitude Networks: HAD Architectures, Methods, and Challenges

This article reviews Hybrid Analog-Digital (HAD) architectures and their associated DOA estimation methodologies, design tradeoffs, and open challenges to enable robust integrated sensing and communication for low-altitude economy networks.

Original authors: Ye Tian, Tuo Wu, Jintao Wu, He Xu, Yuanjun Shen, Xianfu Lei, Kin-Fai Tong

Published 2026-04-02

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ye Tian, Tuo Wu, Jintao Wu, He Xu, Yuanjun Shen, Xianfu Lei, Kin-Fai Tong

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a bustling city in the sky, filled with drones delivering packages, emergency rescue bots, and flying taxis. This is the Low-Altitude Economy (LAE). For all these machines to fly safely without crashing into each other or losing their connection to the ground, they need to know exactly where everything is and how to "talk" to each other.

This paper is about solving a specific problem: How do we build a "super-ear" for these flying machines that can hear exactly where a sound (or signal) is coming from, without being too heavy, expensive, or power-hungry?

Here is the breakdown using simple analogies:

1. The Problem: The "Too Many Microphones" Dilemma

To know exactly where a sound is coming from, you usually need an array of microphones (antennas).

The Old Way (Fully Digital): Imagine a choir of 100 singers, and you hire a separate sound engineer, a high-end microphone, and a massive computer for every single singer. You get perfect sound, but the cost, weight, and electricity bill are astronomical. A small drone simply can't carry this.
The New Way (Hybrid Analog-Digital or HAD): This is the paper's main solution. Imagine the 100 singers are grouped into 10 teams. Each team has a "mixer" (an analog network) that blends their voices together before they reach the sound engineer. Now, you only need 10 engineers and 10 microphones instead of 100.
- The Catch: Because the voices were mixed together, the engineers can't hear the individual singers as clearly. It's like trying to guess who is speaking in a crowded room where everyone is whispering into a single cup. The signal is "compressed," and standard tricks to find the direction of a sound don't work anymore.

2. The Toolkit: Different Types of "Mixers"

The paper looks at different ways to build these mixers (architectures) for different jobs:

The "All-Access" Mixer (Fully Connected): Every singer connects to every engineer. It's the most powerful and precise, like a high-end recording studio. It's great for a big base station on the ground, but too heavy for a drone.
The "Grouped" Mixer (Partially Connected): Singers are locked into specific groups. It's lighter and cheaper, perfect for drones, but you lose a bit of precision.
The "Switchboard" Mixer (Switches-Based): Instead of mixing, you just quickly switch which singers talk to the engineer. It's fast and cheap, good for a quick "hello" to find a signal, but not great for detailed listening.
The "Smart" Mixer (Hybrid Dynamic): This is the best of both worlds. It can switch between being a "Grouped" mixer (for speed) and an "All-Access" mixer (for precision) depending on what the drone is doing at that moment.

3. The Solution: Reconstructing the Puzzle

Since the "mixers" scramble the data, the paper proposes three main ways to unscramble it and find the direction (DOA - Direction of Arrival):

The "Puzzle Reconstructor" (SCM Reconstruction):
Imagine you have a jigsaw puzzle, but someone has glued the pieces together in a weird way. Instead of trying to force the pieces apart, you use math to predict what the original picture must have looked like based on the glued chunks. The paper shows that by mathematically "reconstructing" the full picture from the mixed data, you can get almost as good a result as the expensive "100 microphones" system, but with only 10.
- Key Finding: You don't need more expensive hardware; you just need smarter math.
The "Flashlight Sweep" (Beamforming Scanning):
Instead of trying to see everything at once, you shine a flashlight (a beam) in different directions quickly.
- Step 1: Shine a wide, fuzzy light to find the general area (Coarse).
- Step 2: Once you see something, switch to a narrow, sharp laser to pinpoint it (Fine).
  This is like searching for a lost key in a dark room: first sweep the whole room with a lantern, then use a magnifying glass once you see a glint.
The "Echo Game" (Pilot-Aided):
The sender shouts a specific code (a pilot signal) while the receiver changes its "mixer" settings rapidly. By listening to how the echo changes with each setting, the receiver can build a 3D map of where the sound is coming from, even with limited equipment.

4. Why This Matters for the Future

The paper argues that for the "Sky City" to work, we can't just copy-paste the technology used for ground cell towers.

Reliability is King: In the sky, it's okay if your GPS is slightly off, but it's not okay if your drone loses its connection for a second and crashes. The new methods focus on keeping the connection stable even when buildings block the signal or the drone is moving fast.
The "Swarm" Problem: If 50 drones are flying close together, they might confuse each other's signals. The paper suggests that these drones should talk to each other to share what they "hear," acting like a team of detectives solving a case together rather than working in isolation.

The Bottom Line

This paper is a blueprint for building lightweight, smart, and efficient "ears" for the future of flying technology. It proves that by using clever mixing hardware and advanced math (specifically, reconstructing the signal puzzle), we can give drones and low-flying aircraft the super-power of precise direction-finding without needing to carry a supercomputer on board. It's the difference between a drone that crashes because it got lost, and a drone that navigates a busy city sky with the precision of a human pilot.

1. Problem Statement

The rapid expansion of the Low-Altitude Economy (LAE) (e.g., drones, logistics, emergency response) necessitates robust Direction-of-Arrival (DOA) estimation for Integrated Sensing and Communication (ISAC) systems. Accurate DOA is critical for:

Beam Alignment: Essential for mmWave/sub-THz links to overcome high path loss.
Spatial Reuse: Managing interference in dense swarm scenarios.
Sensing: Airspace situational awareness, target tracking, and interference localization.

The Core Challenge:
While Fully Digital (FD) architectures offer high-resolution DOA, they are impractical for LAE due to stringent Size, Weight, and Power (SWaP) constraints. FD systems require an RF chain and high-speed ADC per antenna, leading to prohibitive power consumption and baseband processing burdens.
Hybrid Analog-Digital (HAD) architectures reduce hardware complexity by compressing antenna-domain observations through analog combining. However, this compression fundamentally alters the measurement model, rendering conventional FD DOA algorithms (like standard MUSIC) ineffective or inaccurate, especially at low Signal-to-Noise Ratios (SNR) or with few RF chains.

2. Methodology & Architectural Framework

The paper proposes a structured approach to HAD-enabled DOA estimation, categorized by architecture and algorithmic strategy.

A. HAD Architectures for LAE

The authors analyze four representative HAD topologies and their suitability for different LAE roles:

Fully Connected (FC-HAD): Connects every RF chain to all antennas via a dense phase-shifter network.
- Pros: High angular resolution, flexible beam shaping.
- Cons: High power/cost, complex calibration.
- Best Fit: Infrastructure (Base Stations) and sensing nodes.
Partially Connected (PC-HAD): Each RF chain connects to a dedicated subarray.
- Pros: Low power, simple routing.
- Cons: Reduced combining freedom, lower resolution.
- Best Fit: Airborne terminals and compact platforms.
Switch-Based (SE-HAD): Uses RF switches to select antenna subsets.
- Pros: Very low loss, fast reconfiguration.
- Cons: Limited instantaneous aperture, coarse resolution.
- Best Fit: Initial access, event-triggered sensing, and portable UE.
Hybrid Dynamic Subarray (HDS): A dynamic bridge between PC and FC, using switches to connect RF chains to subarrays with tunable "closure ratios."
- Pros: Adaptable between low-power (PC-like) and high-resolution (FC-like) modes.
- Best Fit: Dynamic environments requiring a balance of speed and resolution.

B. DOA Estimation Methodologies

The paper reviews three primary algorithmic families designed to overcome HAD compression:

Spatial Covariance Matrix (SCM) Reconstruction:
- Concept: Reconstructs the full-dimensional SCM from compressed hybrid measurements to enable standard subspace methods (e.g., MUSIC).
- Techniques:
  - Element-wise: Solves for unknown covariance entries using multiple combiner configurations.
  - Beamforming-based: Reconstructs covariance blocks by scanning predefined beams.
- Optimization: Leverages the Toeplitz structure of Uniform Linear Arrays (ULA) to reduce the number of parameters to estimate, improving stability and reducing training overhead.
Beamforming-Based Scanning (Coarse-to-Fine):
- Concept: Avoids full SCM reconstruction for speed. Uses a two-stage process:
  1. Coarse Scan: Rapidly scans spatial sectors using wide beams or DFT-based combiners.
  2. Refinement: Performs local fine-sweeping in candidate regions.
- Benefit: Ideal for airborne nodes requiring fast acquisition and re-acquisition after blockage.
Pilot-Aided Estimation:
- Concept: Exploits time-division multiplexing. The transmitter sends multiple pilot sequences while the receiver cycles through different analog combiners.
- Mechanism: Aggregates low-dimensional measurements to form a virtual high-dimensional array, improving accuracy without adding physical RF chains.

3. Key Contributions

Systematic Taxonomy: Provides a comprehensive mapping of HAD architectures (FC, PC, SE, HDS) to specific LAE deployment scenarios (BS, Airborne, UE/GCS), highlighting trade-offs in resolution, power, and robustness.
Algorithmic Review: Synthesizes state-of-the-art HAD-DOA methods, emphasizing SCM reconstruction as the key to bridging the performance gap between HAD and FD systems.
Challenge Identification: Identifies critical open problems specific to LAE, including:
- Reliability-Centric Metrics: Shifting focus from static RMSE to dynamic metrics like loss-of-lock probability and re-acquisition latency.
- Blockage Handling: Developing pipelines for sudden dominant-path switching in urban canyons.
- Multi-User Coupling: Addressing interference in dense swarms where DOA estimates are no longer independent.
- Cooperative Inference: Enabling lightweight exchange of angular features (not raw data) between heterogeneous nodes.

4. Results & Case Studies

The authors present simulation results using a ULA with $M=64$ antennas and two sources ( $10^\circ, 60^\circ$ ):

Performance Gap: Naive application of FD algorithms to HAD data (HAD-MUSIC) results in severe performance degradation at low SNR.
SCM Reconstruction Efficacy: SCM-MUSIC (reconstructing the covariance matrix first) achieves RMSE performance nearly identical to the FD-MUSIC baseline across the entire SNR range, despite using significantly fewer RF chains.
RF Chain Sensitivity: HAD-MUSIC performance degrades sharply as the number of RF chains decreases. In contrast, SCM-MUSIC remains stable.
Design Insight: Increasing the physical array aperture (antenna count) yields greater DOA accuracy gains than increasing the number of RF chains. This validates the strategy of using large, sparse arrays with few RF chains for SWaP-constrained LAE nodes.

5. Significance

This paper is significant because it moves beyond theoretical HAD concepts to address the practical deployment of DOA estimation in the emerging Low-Altitude Economy.

Enabling ISAC: It demonstrates that high-precision sensing and communication can coexist on power-constrained platforms (drones, portable GCS) without the prohibitive cost of fully digital arrays.
Design Guidance: It offers concrete guidelines for system co-design, suggesting that investing in larger physical apertures with SCM reconstruction algorithms is a more viable path for LAE than scaling up RF chains.
Future Roadmap: It sets a research agenda focused on reliability, dynamic blockage handling, and cooperative inference, which are essential for the safety and operational continuity of future low-altitude networks.