2D-DOA Estimation in Switching UCA Using Deep Learning-Based Covariance Matrix Completion.

Area of Science:

• Signal Processing
• Array Signal Processing
• Wireless Communications

Background:

• Direction of Arrival (DOA) estimation is crucial for wireless systems.
• Traditional methods using fully sampled arrays require numerous Radio Frequency (RF) chains, increasing cost and complexity.
• Switching Uniform Circular Arrays (SUCA) offer a potential solution by reducing RF chain requirements.

Purpose of the Study:

• To develop a covariance matrix completion algorithm for 2D-DOA estimation in SUCA.
• To enable accurate 2D-DOA estimation while significantly reducing the number of required RF chains.
• To evaluate the performance of the proposed method against traditional fully sampled arrays.

Main Methods:

• A neural network is employed to estimate the complete covariance matrix of a Fully Sampled Uniform Circular Array (FUCA) from the SUCA's sample covariance matrix.
• The MUSIC algorithm is then applied to the completed covariance matrix for 2D-DOA estimation.
• Monte Carlo simulations are used to assess performance under varying Signal-to-Noise Ratio (SNR) and snapshot counts.

Main Results:

• The proposed algorithm demonstrates that SUCA performance approaches FUCA performance as SNR and snapshot numbers increase.
• The method successfully preserves the advantages of FUCA with a reduced number of RF chains.
• The algorithm is capable of implementing underdetermined 2D-DOA estimation.

Conclusions:

• The developed covariance matrix completion algorithm effectively addresses 2D-DOA estimation challenges in SUCA.
• This approach offers a practical and cost-effective solution for high-performance DOA estimation in wireless systems.
• The algorithm maintains accuracy and enables underdetermined scenarios, highlighting its versatility.