Cyclostationary Signature Embedding

Unlike naturally occurring cyclostationarity from modulation or framing, embedded signatures are synthetically generated by subtly modulating the amplitude, phase, or frequency of the carrier at a specific cyclic frequency (alpha). This creates a controlled spectral correlation peak that is statistically distinguishable from the signal's inherent features. The embedding power is typically 20-30 dB below the data signal, ensuring minimal impact on bit error rate (BER) while remaining detectable by a correlative receiver.

In dynamic spectrum access networks, embedded signatures function as a physical layer rendezvous mechanism. A secondary user can detect the signature of a primary or coordinating node without decoding the full waveform, enabling rapid network discovery. Key applications include:

• Neighbor discovery in ad-hoc cognitive networks
• Spectrum etiquette enforcement by identifying licensed incumbents
• Handoff coordination by signaling a node's presence on a new channel

An embedded cyclostationary signature acts as a watermark for transmitter identity. By assigning a unique cyclic frequency or phase pattern to each device, a receiver can authenticate the source at the physical layer before any cryptographic handshake occurs. This provides defense against replay attacks and MAC address spoofing, as the signature is inseparable from the analog waveform and cannot be stripped by a simple relay.

The signature must be designed to be orthogonal to the host signal's natural cyclostationary features to avoid mutual interference. Common embedding strategies include:

• Amplitude modulation at a sub-harmonic of the symbol rate
• Phase dithering with a known pseudo-random sequence
• Pilot pattern manipulation in OFDM frames Detection is performed using a cyclic feature detector tuned to the known alpha, which correlates the signal with a frequency-shifted version of itself. The processing gain allows detection even when the signature power is far below the noise floor.

Embedded signatures exhibit inherent resilience to multipath fading and Doppler shift because cyclostationary features are preserved through linear time-invariant channels. The cyclic frequency alpha remains constant regardless of the channel's impulse response. For mobile environments, wide-sense cyclostationary signatures can be designed with a spread cyclic period to accommodate Doppler-induced smearing, ensuring reliable detection at vehicular speeds.

Multiple devices can share the same spectrum by embedding orthogonal signatures at distinct cyclic frequencies. A receiver equipped with a bank of cyclic feature detectors can simultaneously identify and separate all active transmitters. This enables code-free multiple access where the signature itself serves as the user identifier, simplifying the MAC protocol and reducing overhead in dense IoT deployments.

The foundational two-dimensional transform for analyzing embedded signatures. The SCF measures the spectral correlation density, revealing the hidden periodicities intentionally inserted into a waveform. An embedded signature manifests as a unique, non-random peak in the SCF at a specific cyclic frequency (alpha) and spectral frequency (f). This representation is the primary domain where a covert identifier is validated against a known template.

A deterministic method for signature embedding that leverages existing communication structures. By subtly altering the phase or amplitude of regularly inserted pilot symbols, a unique periodic statistical pattern is created. This approach is power-efficient because it piggybacks on necessary overhead rather than injecting a separate, energy-wasting signal. The resulting cyclostationary feature is directly tied to the frame structure and can be extracted using a cyclic autocorrelation function (CAF).

A compact, one-dimensional feature vector used to represent an embedded signature for machine learning classifiers. The CDP is generated by projecting the magnitude of the Spectral Correlation Function along the cyclic frequency axis. An intentionally embedded signature creates a distinct, high-magnitude peak in the CDP at the designer-chosen cyclic frequency. This compression makes it ideal for low-latency authentication in cognitive radio nodes.

A FREquency-SHift filtering technique that exploits the known cyclostationarity of an embedded signature to extract it from heavy interference. A FRESH filter is an optimal linear estimator for cyclostationary signals. If a receiver knows the target cyclic frequency of the embedded identifier, it can construct a FRESH filter to isolate that specific signature, even when the signal is buried below the noise floor or masked by a spectrally overlapping interferer.

A robust embedding strategy that places the signature in a higher-order statistical domain to achieve immunity to Gaussian noise. By subtly manipulating the signal's modulation to generate a specific, non-zero cyclic cumulant value, the identifier becomes invisible to second-order (power spectrum) analysis. Extraction requires computing the cyclic polyspectrum, ensuring the signature is only detectable by receivers with the correct higher-order processing capability, adding a layer of covertness.

A statistical hypothesis test used at the receiver to confirm the presence of an embedded signature. The test evaluates the consistency of the cyclic autocorrelation estimate at the pre-shared candidate cyclic frequency. A positive test result rejects the null hypothesis of stationarity, confirming that the intentional periodic pattern is present. This provides a binary authentication gate: the device is verified only if the embedded cyclostationarity is statistically significant.