Radar Constant-Modulus Waveform Design with Prior Information of the Extended Target and Clutter

Wenzhen Yue, Yan Zhang, Yimin Liu, Jingwen Xie2016 · DOI

TL;DR

This paper tackles the challenge of designing radar waveforms that are both constant-modulus (for efficient hardware) and optimized for detecting extended targets in the presence of clutter. They introduce two iterative algorithms that leverage semi-definite relaxation and a novel randomization approach to achieve this, demonstrating improved Signal-to-Clutter-plus-Noise Ratio (SCNR) compared to existing methods.

What Problem Does This Paper Address?

Radar systems rely heavily on waveform design to achieve optimal performance in tasks like target detection, clutter suppression, and identification. A crucial constraint in practical radar systems is the "constant-modulus" property, which simplifies hardware implementation and allows for full transmitter power utilization. However, designing waveforms that satisfy this constraint while also being effective for detecting extended targets (which occupy more than a single resolution cell) in the presence of clutter (unwanted radar returns) is a complex problem. Traditional methods often assume point targets or allow for arbitrary modulus waveforms, limiting their applicability. This paper aims to bridge this gap by developing constant-modulus waveform design techniques that explicitly account for extended targets and clutter, using prior information about them.

Methods

The core of the paper lies in two iterative algorithms for designing constant-modulus waveforms:

  • Arbitrary-Phase Waveform Design: This method uses an iterative approach that alternates between optimizing the receiving filter and the transmitted waveform. To handle the non-convexity of the constant-modulus constraint, they employ semi-definite relaxation (SDR). This transforms the intractable problem into a convex one that can be solved efficiently. A key innovation is a customized randomization approach to recover a valid constant-modulus signal from the relaxed problem's solution.
  • Quadrature Phase Shift Keying (QPSK) Waveform Design: This algorithm is similar to the arbitrary-phase method but specifically designs QPSK waveforms. QPSK signals have only four possible phase states, making them easier to implement than arbitrary-phase waveforms. The paper leverages a technique that shows QPSK signals can approximate a given covariance matrix, allowing for efficient generation of candidate QPSK waveforms.

Both algorithms optimize the Signal-to-Clutter-plus-Noise Ratio (SCNR) by iteratively improving the transmitted waveform and receiving filter.

Datasets

Not discussed in the paper. The paper focuses on theoretical development and simulation-based evaluation using synthetic data generated according to the described signal model.

Key Insights and Ideas

  • The Power of Semi-Definite Relaxation (SDR): SDR is a powerful tool for transforming complex, non-convex optimization problems (like constant-modulus waveform design) into convex ones, making them solvable with standard optimization techniques.
  • Bridging the Gap between Theory and Practice: The constant-modulus constraint is essential for practical radar hardware. This work demonstrates how to achieve near-optimal performance while adhering to this constraint, which is a significant step towards real-world implementation.
  • QPSK as a Practical Alternative: Designing QPSK waveforms offers a compelling trade-off between performance and implementation complexity, making it a valuable option for many radar applications.
  • Importance of Prior Information: The effectiveness of the proposed methods hinges on having prior information about the target and clutter characteristics. This highlights the growing importance of intelligent waveform design that adapts to the environment.
  • Iterative Optimization is Key: The iterative nature of the algorithms allows for gradual improvement of the SCNR, demonstrating that even with constraints, significant performance gains can be achieved.

Key Results

  • The proposed algorithms (Algorithm 1 for arbitrary-phase and Algorithm 2 for QPSK) achieve significant improvements in SCNR compared to existing methods, particularly in the presence of clutter.
  • Algorithm 1 (arbitrary-phase) shows a performance loss of only 0.28 dB compared to a method without the constant-modulus constraint, demonstrating the effectiveness of the SDR and randomization approach.
  • Algorithm 2 (QPSK) offers improved implementation complexity with a slightly larger SCNR loss (0.61 dB) compared to Algorithm 1.
  • In the deterministic target case, the proposed algorithms perform well and are more robust than older iterative methods that can suffer from performance degradation.
  • The QPSK signal design scheme is shown to approximate the original target covariance matrix closely, contributing to its good performance.
  • The proposed techniques can achieve approximately a 5 dB SCNR gain compared to a standard Linear Frequency Modulation (LFM) signal in certain scenarios.

Limitations and Future Work

  • Local Optima: The iterative nature of the alternate optimization strategy means the algorithms might get trapped in local optima, rather than finding the global optimum. The authors suggest that exploring better optimization strategies is an area for future research.
  • Prior Information Estimation: The effectiveness of the proposed methods relies on knowing the statistical characteristics of the target and clutter. Future work could focus on robust methods for estimating this prior information from radar returns.
  • Range Resolution: The current waveform design does not explicitly optimize for range resolution. Future investigations will consider incorporating range resolution as a design objective.
  • Moving Target Case: While the paper mentions that the approach could be extended to moving targets, this is not explored in detail and is suggested for future work.
  • MIMO Radar Extension: The paper focuses on Single-Input Single-Output (SISO) radar, noting that extension to Multiple-Input Multiple-Output (MIMO) radar is straightforward and a potential future direction.

Summary updated

BibTeX
@misc{yue2016radar,
  author = {Wenzhen Yue and Yan Zhang and Yimin Liu and Jingwen Xie},
  title = {{Radar Constant-Modulus Waveform Design with Prior Information of the Extended Target and Clutter}},
  year = {2016},
  doi = {10.3390/s16060889},
}
Figure