Missed Detection Probability

A primary cause of missed detection where the sensing node is physically shadowed or in a deep fade relative to the primary transmitter. The sensor perceives a vacant channel while a primary receiver nearby is actively receiving.

• Multipath Fading: Destructive interference causes deep signal nulls at the sensor's location.
• Shadowing: Large obstacles like buildings or terrain block the line-of-sight path.
• Mitigation: Requires cooperative spectrum sensing with spatially diverse nodes to overcome local fading effects.

The inherent fluctuation in ambient noise power fundamentally limits non-coherent detectors like the energy detector. Noise uncertainty creates a Signal-to-Noise Ratio (SNR) Wall: a threshold below which no amount of sensing time can reliably distinguish a signal from noise.

• Source: Thermal noise variations, amplifier gain fluctuations, and environmental interference.
• Impact: An energy detector operating below the SNR Wall will suffer a high missed detection probability regardless of the threshold setting.
• Countermeasure: Cyclostationary feature detection exploits signal periodicity to bypass the SNR Wall.

A fundamental design constraint in cognitive radio frame structure. Increasing the sensing duration improves detection accuracy but reduces the time available for data transmission, lowering throughput.

• Frame Structure: A MAC frame is divided into a sensing slot and a transmission slot.
• Trade-off Dynamics: A longer sensing slot reduces missed detection probability but increases the false alarm probability for a fixed threshold, or vice versa.
• Optimization: Deep Reinforcement Learning agents can dynamically adjust the sensing duration to find the Pareto-optimal balance based on real-time channel conditions.

Spectrum sensing typically focuses on detecting the primary transmitter, but the interference actually occurs at the primary receiver. The sensor may not know the receiver's location or its local noise floor.

• Duplex Distance: The physical separation between the primary transmitter and its intended receiver means the sensor's measurement is not a direct proxy for interference potential.
• Local Interference: A primary receiver may experience high local interference that is invisible to the remote sensor.
• Solution: Radio Environment Maps (REMs) integrate geolocation and propagation models to estimate the interference power at the protected receiver.

A malicious security threat where an attacker mimics the spectral characteristics of a legitimate primary user. This can force a cognitive radio into a state of confusion, potentially causing it to misclassify a real primary signal or vacate a channel unnecessarily.

• Mechanism: The attacker transmits a signal with the known modulation, pilot patterns, or cyclostationary signatures of the primary user.
• Consequence: A sensor trained on these features may fail to detect the real primary user if the attacker's signal saturates the detection algorithm.
• Defense: Radio Frequency Fingerprinting uses deep learning to identify unique hardware-level imperfections that cannot be easily emulated.

The selection of the detection threshold directly controls the trade-off between missed detection and false alarm probabilities, as visualized by the Receiver Operating Characteristic (ROC) curve.

• Constant False Alarm Rate (CFAR): Adapts the threshold to maintain a fixed false alarm rate, but the resulting missed detection probability will vary with the instantaneous SNR.
• Neyman-Pearson Criterion: Fixes the false alarm probability and maximizes the detection probability, but requires accurate noise power estimation.
• Bayesian Approach: Minimizes the total error probability by incorporating prior probabilities of channel occupancy, but is sensitive to inaccurate priors.

The conditional probability that a sensing algorithm incorrectly declares a frequency band as occupied when it is actually vacant. This error represents a lost transmission opportunity for the secondary user. It is the direct statistical counterpart to Missed Detection Probability, and the two are traded off against each other via the detection threshold. A system optimized solely to minimize missed detections will inevitably suffer from a high false alarm rate, crippling spectral efficiency.

A graphical plot that illustrates the diagnostic ability of a binary classifier as its discrimination threshold is varied. The curve plots the Probability of Detection (Pd) against the False Alarm Probability (Pfa). The ROC curve is the primary tool for visualizing the inherent trade-off between missed detections and false alarms. The area under the curve (AUC) quantifies overall sensing efficacy; a perfect sensor achieves an AUC of 1.0, while a random guess yields 0.5.

A degradation in sensing reliability caused when a cognitive radio is shadowed or in a deep fade relative to a transmitting primary user. The sensor cannot detect the primary user's signal due to physical obstruction, leading directly to a missed detection and subsequent harmful interference when the secondary user begins transmitting. This problem is the primary motivation for cooperative sensing architectures that aggregate observations from spatially diverse nodes.

The theoretical minimum SNR threshold below which a non-coherent detector cannot reliably distinguish a signal from noise, regardless of how long it observes the spectrum. This fundamental limit arises from noise uncertainty—the inherent fluctuation in ambient noise power. Below the SNR wall, the missed detection and false alarm probabilities both approach 0.5, rendering the sensor useless. This concept defines the absolute sensitivity boundary for energy detection.

A distributed architecture where multiple cognitive radios share local sensing observations to mitigate the hidden node problem and improve overall detection reliability. By combining data from spatially diverse sensors, the network dramatically reduces the aggregate Missed Detection Probability. Fusion strategies include:

• Hard Decision Fusion: Nodes transmit binary busy/idle votes to a fusion center
• Soft Decision Fusion: Nodes share raw energy levels or likelihood ratios for superior statistical combining

An adaptive threshold-setting algorithm that maintains a fixed, predetermined False Alarm Probability despite variations in background noise power. CFAR processors continuously estimate the local noise floor from neighboring cells and dynamically adjust the detection threshold. This ensures that the Missed Detection Probability is the only variable that fluctuates with signal conditions, providing predictable and manageable interference risk for spectrum regulators.