Compressive Spectrum Sensing

The foundational principle that allows sampling far below the Nyquist rate. Wideband spectrum is inherently sparse; most frequencies are unoccupied. Compressive sensing exploits this by acquiring a small number of linear, non-adaptive measurements that capture the essential information of the signal. Instead of sampling uniformly in time, the system projects the high-dimensional signal onto a lower-dimensional basis, enabling Analog-to-Information Conversion (AIC). This dramatically reduces the burden on the Analog-to-Digital Converter (ADC) and subsequent digital processing chains.

The algorithmic recovery of the original high-resolution spectrum from the compressed measurements. Since the system is underdetermined, traditional linear methods fail. Recovery relies on convex optimization techniques that seek the sparsest solution. Common algorithms include:

• L1-Minimization (Basis Pursuit): Minimizes the L1-norm of the signal coefficients, promoting sparsity.
• Greedy Pursuits (OMP, CoSaMP): Iteratively select the most significant spectral components. This reconstruction is the computational cost traded for the hardware savings.

The mathematical condition that guarantees stable and robust signal recovery. A measurement matrix satisfies the RIP if it approximately preserves the Euclidean length of all sufficiently sparse vectors. In essence, it ensures that distinct sparse signals remain distinct after compression. While computationally difficult to verify for a specific matrix, random matrices satisfy the RIP with overwhelming probability when the number of measurements M scales as M ≥ C * K * log(N/K), where K is the sparsity level and N is the ambient dimension. This provides the theoretical performance guarantee for the entire system.

The foundational signal acquisition method that enables compressive sensing by sampling below the Nyquist rate. It exploits the sparse structure of the signal in a specific domain—such as frequency—to reconstruct the original wideband signal from far fewer samples than traditional Nyquist-Shannon theory requires.

• Key mechanism: Uses random demodulation or multi-coset sampling architectures
• Hardware benefit: Replaces expensive high-rate ADCs with multiple low-rate samplers
• Reconstruction requirement: Requires non-linear optimization or greedy algorithms like Orthogonal Matching Pursuit (OMP)

The process of simultaneously monitoring a broad, contiguous block of frequencies to identify multiple spectrum holes across a wide range. Traditional approaches require Nyquist-rate ADCs that are power-hungry and expensive, making compressive techniques essential for practical deployment.

• Challenge: GHz-wide bandwidths demand sampling rates exceeding tens of GS/s
• Compressive solution: Leverages spectral sparsity to reduce the sampling burden
• Application: Critical for cognitive radios operating in spectrum-rich environments like radar and 5G bands

A frequency band assigned to a primary user that is unoccupied at a specific time and geographic location. Compressive sensing accelerates the identification of these opportunities by rapidly scanning wide bandwidths for spectral vacancies without requiring per-channel sequential scanning.

• Temporal dimension: Holes appear and disappear dynamically as primary users activate
• Spatial dimension: A hole at one location may be occupied at another
• Detection goal: Maximize probability of detection while minimizing false alarms

A blind spectrum sensing technique that compares received signal energy against a noise-dependent threshold. When combined with compressive sampling, energy detection can be performed directly on compressed measurements without full signal reconstruction, enabling fast, low-complexity occupancy decisions.

• Advantage: No prior knowledge of primary user signal required
• Limitation: Performance degrades severely under noise uncertainty
• Compressive variant: Compressed energy detection operates on sub-Nyquist samples directly

The theoretical minimum SNR threshold below which a non-coherent detector cannot reliably distinguish a signal from noise, regardless of observation time. Compressive sensing architectures must account for this fundamental limit, as noise uncertainty creates an irreducible performance floor.

• Cause: Inherent fluctuation in ambient noise power estimation
• Impact: Creates a detection dead zone even with infinite samples
• Mitigation: Cyclostationary feature detection or eigenvalue-based methods can push below the SNR wall

A robust sensing method that exploits the periodic statistical properties of modulated signals to distinguish them from stationary noise. When integrated with compressive architectures, it offers superior low-SNR performance compared to energy detection, though at higher computational cost.

• Mechanism: Detects spectral correlation at cyclic frequencies unique to each modulation scheme
• Advantage: Resilient to noise uncertainty and can identify signal type
• Compressive challenge: Requires careful design to preserve cyclostationary features in compressed measurements