Double-Layer Compressive Sensing Based Efficient DOA Estimation in WSAN with Block Data Loss

Peng Sun, Liantao Wu, Kai Yu, Huajie Shao, Zhi Wang

TL;DR

This paper tackles the critical problem of Direction of Arrival (DOA) estimation in Wireless Sensor Array Networks (WSANs) where data loss, particularly block data loss due to large packet sizes, severely degrades performance. They introduce a novel "double-layer compressive sensing" framework that combines passive modeling of packet loss with an active sensing step at each sensor to mitigate these issues. This framework enables both a two-stage (DCS-DOA) and a direct (DCS-DDOA) DOA estimation approach, both demonstrating significant improvements over existing methods in the presence of data loss.

What Problem Does This Paper Address?

Accurate estimation of the Direction of Arrival (DOA) is crucial for many applications, including smart antennas, mobile communications, and target tracking. In Wireless Sensor Array Networks (WSANs), sensors communicate wirelessly with a central Fusion Center (FC). However, low-power wireless links are inherently unreliable, leading to packet loss. This problem is exacerbated when large packet sizes are used to improve transmission efficiency, as losing a single large packet results in significant "block data loss." This block data loss can catastrophically degrade the performance of traditional DOA estimation techniques. Existing methods often focus on signal recovery rather than the specific application of DOA estimation in such lossy environments.

Methods

The core contribution is a double-layer compressive sensing (CS) framework. This framework addresses block data loss in two key ways:

  1. Passive CS Process: Random packet loss during transmission is modeled as a passive CS process. This essentially means that the received data at the FC is a sparse subset of the transmitted data, governed by the packet reception rate and packet size.
  2. Active CS Process: An active CS procedure is introduced at each array sensor before transmission. This involves a dimension-reduced projection of the raw sensor data using a Gaussian or permutation matrix. This projection ensures that each transmitted data sample is a weighted average of the original signals, making the data more robust to loss. It also reduces the data volume, mitigating the impact of packet loss.

Under this framework, two DOA estimation techniques are proposed:

  • DCS-DOA (Double-Layer CS - DOA): This is a two-stage approach. First, the raw sparse vectors are recovered from the received (potentially incomplete) data at the FC using sparsity-based techniques. Second, DOA estimation is performed on these recovered vectors.
  • DCS-DDOA (Double-Layer CS - Direct DOA): This method aims to avoid the error propagation inherent in the signal recovery stage of DCS-DOA. It leverages a joint frequency and spatial domain sparse representation of the sensor array data. This allows the FC to directly estimate DOAs from the received data packets, skipping the signal recovery step.

Datasets

Not discussed in the paper. The authors use synthesized data for simulations.

Key Insights and Ideas

  • Block Data Loss is a Major Hurdle: The paper clearly demonstrates how increasing packet size, while improving efficiency in ideal conditions, significantly amplifies the negative impact of block data loss on DOA estimation accuracy.
  • Double-Layer CS is Key: The combination of modeling packet loss passively and actively projecting data at the sensor is crucial for achieving robust DOA estimation. The active CS step effectively "scatters" information across packets, making individual packet losses less detrimental.
  • Direct DOA Estimation Avoids Error Propagation: The DCS-DDOA approach, by directly estimating DOAs from the received measurements without an intermediate signal recovery step, bypasses a significant source of error, leading to improved performance.
  • Joint Frequency and Spatial Sparsity: Exploiting the inherent sparsity in both the frequency and spatial domains is a powerful technique for direct DOA estimation in this context.

Key Results

  • Significant Improvement over CS-DOA: Both DCS-DOA and DCS-DDOA consistently outperform a baseline single-layer CS-DOA method, especially under conditions of high packet loss and larger packet sizes.
  • DCS-DDOA Superiority: DCS-DDOA generally shows better performance (lower RMSE, higher detection frequency) than DCS-DOA, particularly in challenging scenarios, due to the avoidance of error propagation.
  • Robustness to Packet Size: The double-layer CS framework makes DOA estimation performance largely independent of packet size, allowing for the use of larger packets for improved transmission efficiency without sacrificing accuracy.
  • Performance Trade-offs: While DCS-DDOA offers superior accuracy, it incurs higher computational time compared to DCS-DOA. The choice between them depends on the real-time requirements and link quality.

Limitations and Future Work

  • Computational Cost of DCS-DDOA: The paper acknowledges that DCS-DDOA requires more computation time than DCS-DOA, making it less suitable for applications with extremely strict real-time constraints.
  • Link Quality Estimation: The authors suggest that if link quality can be estimated in real-time, a hybrid approach could be adopted, choosing DCS-DOA for good links and DCS-DDOA for poor links. This implies future work in real-time link quality assessment.
  • Generalizability: While the paper focuses on Uniform Linear Arrays (ULAs), extending the framework to non-uniform arrays or different sensor configurations could be a future direction.
  • Real-world Deployment: The simulations are based on synthesized data. Further validation with real-world WSAN deployments would be valuable.

Summary updated

Abstract

Accurate information acquisition is of vital importance for wireless sensor array network (WSAN) direction of arrival (DOA) estimation. However, due to the lossy nature of low-power wireless links, data loss, especially block data loss induced by adopting a large packet size, has a catastrophic effect on DOA estimation performance in WSAN. In this paper, we propose a double-layer compressive sensing (CS) framework to eliminate the hazards of block data loss, to achieve high accuracy and efficient DOA estimation. In addition to modeling the random packet loss during transmission as a passive CS process, an active CS procedure is introduced at each array sensor to further enhance the robustness of transmission. Furthermore, to avoid the error propagation from signal recovery to DOA estimation in conventional methods, we propose a direct DOA estimation technique under the double-layer CS framework. Leveraging a joint frequency and spatial domain sparse representation of the sensor array data, the fusion center (FC) can directly obtain the DOA estimation results according to the received data packets, skipping the phase of signal recovery. Extensive simulations demonstrate that the double-layer CS framework can eliminate the adverse effects induced by block data loss and yield a superior DOA estimation performance in WSAN.
BibTeX
@inproceedings{sun2017double,
  author = {Peng Sun and Liantao Wu and Kai Yu and Huajie Shao and Zhi Wang},
  title = {{Double-Layer Compressive Sensing Based Efficient DOA Estimation in WSAN with Block Data Loss}},
  booktitle = {Sensors},
  year = {2017},
  url = {https://openreview.net/forum?id=pEvDnRjSPH},
}
Figure