Spectrum Sensing

A non-coherent detection method that compares the received signal energy against a noise-dependent threshold. It is the most computationally efficient technique, requiring no prior knowledge of the primary user's signal structure.

• Mechanism: Squares and integrates the received signal over time; if the output exceeds a threshold, the channel is declared occupied.
• Key Limitation: Susceptible to noise uncertainty—small fluctuations in the noise floor dramatically degrade performance, creating a SNR wall below which detection becomes impossible.
• Use Case: Ideal for rapid, coarse sensing in high-SNR environments where power consumption is the primary constraint.

A coherent detection technique that maximizes the output signal-to-noise ratio by correlating the received signal with a known replica of the primary user's waveform. It is the optimal detector when full signal knowledge is available.

• Mechanism: Convolves the incoming signal with a time-reversed template of the known pilot, preamble, or spreading code.
• Key Trade-off: Requires perfect synchronization and demodulation of the primary signal, demanding a dedicated receiver for each modulation type.
• Use Case: Deployed when secondary systems have access to standardized pilot structures, such as ATSC digital TV or LTE cell-specific reference signals.

An advanced technique that exploits the periodic statistical properties inherent in modulated signals to distinguish them from stationary noise. It is robust to noise uncertainty and can identify the specific modulation type.

• Mechanism: Computes the spectral correlation function (SCF) to detect cyclic frequencies caused by carrier tones, symbol rates, and guard intervals.
• Key Advantage: Can differentiate between primary users, secondary users, and interference by analyzing unique cyclic signatures.
• Use Case: Critical for automatic modulation classification (AMC) and environments where energy detection fails due to unknown or variable noise floors.

A blind sensing method that uses random matrix theory to analyze the eigenvalues of the received signal's sample covariance matrix. It requires no knowledge of signal, noise, or channel characteristics.

• Mechanism: Computes the ratio of the maximum to minimum eigenvalue; if the ratio exceeds a theoretically derived threshold, a signal is present.
• Key Algorithms: Includes the Maximum-Minimum Eigenvalue (MME) detector and the Energy with Minimum Eigenvalue (EME) detector.
• Use Case: Excels in multi-antenna systems and cooperative sensing scenarios where spatial diversity provides a rich covariance structure.

A correlation-based method that exploits known repetitive patterns within the target signal's time-domain structure, such as preambles, cyclic prefixes, or training sequences.

• Mechanism: Performs autocorrelation on the received signal to detect periodicity introduced by the cyclic prefix in OFDM systems or the synchronization preamble in Wi-Fi frames.
• Key Advantage: Converges faster than energy detection and provides reliable performance at lower SNRs when the waveform structure is standardized.
• Use Case: Highly effective for detecting OFDM-based signals like Wi-Fi, LTE, and DVB-T where the cyclic prefix length is known.

A data-driven approach that trains convolutional neural networks (CNNs) or recurrent neural networks (RNNs) on raw IQ samples or spectrograms to perform simultaneous detection and classification without explicit feature engineering.

• Mechanism: A neural network learns hierarchical representations directly from the time-frequency domain, mapping complex signal structures to a binary occupancy decision.
• Key Advantage: Outperforms model-based methods in low-SNR regimes and complex interference environments by learning non-linear decision boundaries.
• Use Case: Deployed in next-generation cognitive radios requiring joint spectrum sensing and automatic modulation classification in dense, heterogeneous spectrum environments.

A spectrum sharing mechanism where unlicensed secondary users autonomously identify and utilize vacant licensed spectrum bands without causing harmful interference to primary users. DSA relies on real-time spectrum sensing to detect spectrum holes and dynamic frequency selection to vacate channels immediately upon primary user return. This is the operational objective that spectrum sensing enables.

A denial-of-service security threat where a malicious actor mimics the signal characteristics of a licensed primary user to prevent legitimate secondary users from accessing vacant spectrum. Advanced detection relies on RF fingerprinting and location verification to distinguish authentic primary signals from spoofed emulations, making robust sensing a critical security layer.

A multi-dimensional database integrating geolocation, spectrum policies, propagation models, and real-time sensing data to provide comprehensive awareness of the RF landscape. REMs enable predictive spectrum sensing by fusing historical occupancy patterns with current measurements, allowing cognitive radios to anticipate channel availability rather than simply reacting to it.

A collaborative detection framework where multiple secondary users share their local sensing observations with a fusion center to improve detection reliability. This approach mitigates the hidden node problem, where a single sensor is physically obstructed from detecting a primary transmitter. Hard combining rules use binary decisions, while soft combining leverages raw energy measurements for superior sensitivity.

The fundamental sensing tradeoff: a false alarm incorrectly declares a band occupied, wasting transmission opportunities, while a missed detection fails to identify an active primary user, causing harmful interference. Receiver operating characteristic curves quantify this balance, and regulatory bodies like the FCC mandate maximum missed detection probabilities to protect incumbent services.

A machine learning technique that autonomously identifies the modulation scheme of a received signal without prior knowledge. AMC serves as an intelligent front-end for spectrum sensing by classifying signal types to distinguish primary users from interferers. Deep learning architectures now achieve high accuracy across diverse modulation formats including QPSK, 16-QAM, and 64-QAM in low SNR conditions.