Abstract
This paper proposes a novel multi-channel compressed structure to estimate directions of arrival (DOAs) of broadband signals based on the Nyquist folding structure. The research first introduces a multi-channel receiving structure on a long-short baseline interferometer array and analyzes its feasibility for broadband DOA estimation. A specialized frequency estimation algorithm is developed for input signals under this architecture. Finally, DOAs are estimated using a long-short baseline interferometer algorithm. Computer simulations demonstrate that the proposed system works effectively, providing stable DOA estimation for broadband signals when the signal-to-noise ratio (SNR) is greater than 0 dB.
Introduction
In modern electronic warfare and reconnaissance, receivers must monitor increasingly wide swaths of the radio frequency (RF) spectrum. Traditional Nyquist sampling requires Analog-to-Digital Converters (ADCs) to sample at twice the highest frequency of interest. For broadband signals spanning several gigahertz (e.g., 6–10 GHz), this requirement leads to prohibitively expensive hardware, extreme data rates, and high power consumption.
Compressed sensing techniques, such as the Nyquist Folding Receiver (NYFR), offer a solution by capturing wideband signals using slower ADCs through intentional spectral folding. However, determining the Direction of Arrival (DOA) becomes complex once phase and frequency information are scrambled by the folding process. This paper addresses high-precision DOA estimation within a sub-Nyquist, multi-channel compressed sampling framework.
Proposed Methodology
Multi-Channel Nyquist Folding Receiver (MCNYFR)
The authors propose the Multi-Channel Nyquist Folding Receiver (MCNYFR), which utilizes a modulated local oscillator (LO) in each channel to "fold" the wide RF input into a narrow baseband. This architecture allows a 1 GHz ADC to effectively capture signals from a 4 GHz wide bandwidth, reducing the data processing load and hardware cost by 75% compared to a full Nyquist system.
Long-Short Baseline Array Configuration
To achieve both unambiguous and high-precision DOA estimation, the system employs an antenna array with a specific long-short baseline configuration. Short baselines (where spacing is less than half a wavelength) provide a coarse, unambiguous estimate, while long baselines (spacing greater than half a wavelength) offer high precision but suffer from phase ambiguity. By integrating these two, the system resolves "ghost" directions while maintaining accuracy.
Phase Preservation and Synchronization
A critical insight of this work is that relative phase differences between channels remain constant despite spectral folding. To exploit this, the digital signal processor (DSP) must be perfectly synchronized with the analog modulation. The researchers implemented a "Synchronous LO Group" using a closed-loop test to compensate for time delays, ensuring phase consistency across the array.
Simulation and Results
The proposed system was validated through computer simulations using a 4-element linear array and a 4 GHz bandwidth (6–10 GHz) sampled at 1 GHz. Key findings include:
- Frequency Estimation: The proposed algorithm significantly outperformed existing methods in noisy environments, particularly when the SNR was below 0 dB.
- DOA Stability: Stable DOA estimation was achieved once the SNR exceeded 0 dB. The Normalized Root Mean Square Error (NRMSE) for DOA estimation decreased sharply as SNR increased, confirming the effectiveness of the long-short baseline approach.
- Efficiency: The results demonstrate that a 1 GHz sampling rate is sufficient to characterize signals across a 4 GHz bandwidth without sacrificing directional precision.
Conclusion
The MCNYFR architecture provides a robust and cost-effective method for broadband DOA estimation. While the current method faces challenges with frequency collisions (where two signals share the exact same carrier frequency), it proves highly effective for single-frequency components of broadband signals. Future work will focus on validating these results with real-world hardware prototypes to account for RF impairments like IQ imbalance and mixer non-linearities.
Summary updated