The compact high-frequency surface wave radar using a crossed-loop/monopole (CLM) antenna as the receiving sensor has been widely used in ocean remote sensing and target monitoring. However, the direction of arrival (DOA) estimation accuracy of a single CLM antenna is the dominant factor that restricts the target monitoring performance of the compact HF radar. Besides, the single CLM antenna can estimate two signals simultaneously at most, but its effectiveness is challenged by the pattern distortion and the existence of coherent sources, which limits the application range of the compact HF radar. In this study, a compact array combining two CLM antennas is proposed to improve the DOA estimation accuracy and solve the multisource DOA estimation problem. The estimation error and multisource DOA estimation performance of a dual CLM antenna array are analyzed by formula derivation and simulation. Furthermore, the field experiment results are given to demonstrate the performance improvement of the dual CLM antenna array.
By using the vertical polarization electromagnetic wave that travels along the curvature of ocean surface with low loss, the high-frequency surface wave radar (HFSWR) can be used for real-time ocean remote sensing over the horizon range, including not only the measurement of ocean surface dynamic parameters [
However, the compact HF radar with the single CLM antenna has some defects in practical application. Firstly, the excessive DOA estimation error caused by antenna pattern distortion in practice significantly reduces the performance of sea-state remote sensing [
An effective way is using a dual CLM antenna array as the receiving sensor to solve the aforementioned problems while retaining the advantage of its small size. In this study, the DOA estimation performance of the dual CLM antenna array is investigated. The estimation error of the dual CLM antenna array based on the MUSIC algorithm is theoretically derived. Then, simulations are conducted to analyze the multisource DOA estimation performance. The estimation results of field experimental data are given to prove that the DOA estimation accuracy of the dual CLM antenna array is higher than that of the single CLM antenna, which is consistent with the theoretical formula. Moreover, a multisource case of vessel targets in the shore-based HF radar data is studied, the result also confirms that it is effective to estimate the DOAs of multisource with dual CLM antenna array.
This article is organized as follows. Section
In the case of single signal source, the signal model can be written as
The steering vector of a CLM antenna with the normal direction of 0° can be written as
The dual CLM antenna array is a combination of uniform linear array (ULA) and CLM antenna, as shown in Figure
Diagram of a dual CLM antenna array, where monopole, loop B, and loop A are the three elements of the single CLM antenna, and
Stoica and Nehorai [
For an idealized CLM antenna, the MUSIC error variance has been deduced by [
Similarly, the MUSIC error variance of the dual CLM antenna array can be expressed as
Comparing (
Theoretical MUSIC error of different antenna types. For the dual CLM array and ULAs, the array spacing is set as half-wavelength.
According to [
To analyze the multisource DOA estimation performance of the dual CLM antenna array under the incoherent and coherent conditions, Monte Carlo simulations are carried out in this section. The simulation parameters are as follows: the number of snapshots is 30, and the snapshots are statistically independent, the number of independent trial times is 500, and the spacing of adjacent elements is half of the wavelength.
Figure
MUSIC spectrum of the dual CLM antenna array versus the SNR with incident angles of 0° and 40°. (a) Incoherent signals. (b) Coherent signals.
In our simulations, we define two signals coming from
Simulation results of the successful estimation probability in incoherent and coherent cases. (a) Probability of successful estimation versus the SNR with incident angles of 0° and 40°. (b) Probability of successful estimation versus incident angles when SNR = 20 dB.
In Figure
In order to validate the experimental performance of the dual CLM antenna array, a field experiment was carried out at Longhai (
The constant false-alarm-rate (CFAR) algorithm [
Figure
DOA estimation results in Longhai. (a) Antenna-1. (b) Antenna-2. (c) Dual antenna array.
Statistics of DOA estimation results.
Antenna-1 | Antenna-2 | Dual antenna array | |
---|---|---|---|
Corr. coef | 0.816 | 0.663 | 0.845 |
RMSE (degree) | 14.96 | 23.51 | 13.27 |
Figure
The error distribution of DOA estimation results.
Figure
Comparison of simulated and experimental DOA estimation errors at different SNRs. The incident azimuth of the signal is −20°.
Figure
A case of multisources with target azimuths of 65° and 177°. (a) RD spectra. (b) Doppler spectra at the 9th range bin (22.5 km). Color bar in (a) is used to illustrate the signal strength in dB.
Figure
Normalized MUSIC spectrum of targets with actual azimuths of 65° and 177°.
In this article, we propose to use the dual CLM antenna array as the receiving sensor of the HF radar to improve the DOA estimation accuracy and solve the multisource DOA estimation problem. The DOA estimation accuracy and multisource estimation performance of the dual CLM antenna array are studied by formula derivation and simulations. The estimation results of thousands of detected samples prove that the dual CLM antenna array has higher estimation accuracy than the single CLM antenna. The estimation error of experimental data is also consistent with the theoretical curve. A multisource case in shore-based radar data is investigated, and the result verifies that the dual CLM antenna array can achieve multisource DOA estimation while the single CLM antenna fails. This improvement extends the application range of the compact HF radar and makes it possible to install the radar in the environment such as the floating platform.
The data used to support the findings of this study are available from the corresponding author upon request.
The authors declare that there are no conflicts of interest regarding the publication of this paper.
This work was supported by the National Natural Science Foundation of China (Grant nos. 61671331, 41706200, and 61661009), the National Key R&D Program of China (Grant no. 2017YFC0405703), the Fundamental Research Funds for the Central Universities (Grant no. 2042018kf1009), and the Natural Science Foundation of Jiangsu Province (Grant no. SBK2019040262).