Blind Sensing

The foundational mechanism of blind sensing. This approach computes the sample covariance matrix of the received signal and performs eigenvalue decomposition. Under the noise-only hypothesis, the eigenvalues are approximately equal; under the signal-present hypothesis, the largest eigenvalue significantly exceeds the others. Common test statistics include the Maximum-to-Minimum Eigenvalue (MME) ratio and the Generalized Likelihood Ratio Test (GLRT) derived from random matrix theory.

A computationally efficient blind detection method that computes the ratio of the sum of absolute values of all elements in the sample covariance matrix to the sum of absolute values of its diagonal elements. The CAV statistic approaches a known theoretical value under the noise-only hypothesis and deviates when a correlated signal is present. This method avoids the computationally expensive eigenvalue decomposition step, making it suitable for resource-constrained cognitive radios.

The critical advantage of blind sensing over energy detection. Energy detectors suffer from a Signal-to-Noise Ratio (SNR) wall below which reliable detection becomes impossible due to imprecise noise power estimation. Blind sensing methods are inherently robust to noise uncertainty because they rely on the correlation structure of the signal rather than absolute energy levels. This makes them effective in low-SNR environments where traditional methods fail catastrophically.

The theoretical backbone of modern blind sensing. When the number of samples and the number of receiving antennas are both large, the empirical distribution of eigenvalues of the sample covariance matrix converges to the Marchenko-Pastur law under the noise-only hypothesis. This allows for precise, analytical threshold setting based on the Tracy-Widom distribution for the largest eigenvalue, enabling constant false alarm rate operation without empirical calibration.

Blind sensing inherently leverages spatial diversity from multiple receiver antennas. The covariance matrix captures cross-correlations between antenna elements, which are present for structured signals but absent for uncorrelated noise. This makes blind sensing particularly effective in MIMO cognitive radio systems where multiple antennas are already available for communication, requiring no additional hardware for sensing.

Unlike matched filter detection or cyclostationary feature detection, blind sensing requires zero a priori knowledge of the primary user's signal. It does not need to know the modulation scheme, carrier frequency offset, symbol rate, pulse shaping filter, or any other waveform-specific parameter. This makes it universally applicable across heterogeneous spectrum environments where multiple primary user types with unknown signal characteristics may operate simultaneously.

A foundational non-coherent method that measures received signal energy over a time-frequency block and compares it to a threshold. It requires no prior knowledge of the primary user's signal structure, making it a classic blind technique.

• Mechanism: Squares and integrates the received samples.
• Key Limitation: Highly susceptible to noise uncertainty, which creates an SNR wall below which detection becomes impossible.
• Application: Often used as a first-pass coarse sensor in hierarchical sensing frameworks.

A class of blind algorithms that exploit the statistical properties of the received signal's sample covariance matrix. These methods derive test statistics from the eigenvalues of the matrix to distinguish signal from noise.

• Maximum-Minimum Eigenvalue (MME): Computes the ratio of the largest to smallest eigenvalue.
• Energy with Minimum Eigenvalue (EME): Uses the ratio of average power to the minimum eigenvalue.
• Advantage: Robust against noise uncertainty because noise power estimation is not required.

A blind detection algorithm that computes a test statistic from the ratio of the sum of absolute values of all elements in the sample covariance matrix to the sum of its diagonal elements.

• Principle: Off-diagonal elements of the covariance matrix are non-zero when a correlated signal is present, but approach zero for uncorrelated white noise.
• Complexity: Lower computational overhead than full eigenvalue decomposition, making it suitable for resource-constrained cognitive radios.
• Performance: Approaches the detection capability of eigenvalue-based methods with reduced processing requirements.

A blind sensing framework that tests whether the empirical distribution of received samples matches a hypothesized noise distribution. No signal structure knowledge is required.

• Anderson-Darling Test: Measures the weighted distance between the empirical cumulative distribution function (CDF) and the theoretical noise CDF.
• Kolmogorov-Smirnov Test: Uses the supremum of the absolute difference between empirical and theoretical CDFs.
• Benefit: Performs well with a small number of samples and is non-parametric in nature.

The fundamental performance limit for blind sensing, particularly energy detection. Noise uncertainty refers to the receiver's inability to precisely estimate the ambient noise power due to thermal fluctuations, component non-linearity, and calibration errors.

• SNR Wall: A signal-to-noise ratio threshold below which no detector can reliably distinguish signal from noise, regardless of sensing duration.
• Mitigation: Eigenvalue-based and cyclostationary methods are inherently robust to noise uncertainty because they do not rely on absolute noise power estimates.
• Practical Impact: Defines the operational boundary for secondary user deployment density.

An adaptive threshold-setting methodology critical for practical blind sensing deployment. CFAR algorithms dynamically adjust the detection threshold to maintain a fixed, pre-defined probability of false alarm despite fluctuating noise floors.

• Cell-Averaging CFAR: Estimates local noise power by averaging neighboring range bins or frequency cells.
• Ordered-Statistic CFAR: Uses the k-th ordered sample from reference cells to set the threshold, offering robustness against interfering signals in the reference window.
• Relevance: Ensures stable regulatory compliance by preventing excessive false alarms that waste spectrum opportunities.