RF Watermarking

The watermark signal is injected at a power level significantly below the noise floor of the primary data channel. A standard receiver demodulates the primary payload with negligible signal-to-noise ratio (SNR) degradation, typically less than 0.5 dB. The watermark is designed to be statistically orthogonal to the modulated data, appearing as benign, uncorrelated noise to legacy hardware. This ensures backward compatibility with existing receiver populations while adding a covert authentication layer detectable only by an intended monitor.

The authentication tag is spread across a wide bandwidth using a secret pseudo-noise (PN) sequence known only to the transmitter and the verifier. This processing gain pushes the watermark's power spectral density far below the thermal noise floor, making it cryptographically invisible to unintended observers. Key characteristics include:

• Low Probability of Intercept (LPI): The signal is statistically indistinguishable from noise without the correct despreading key.
• Interference Rejection: The despreading process at the verifier collapses the watermark while suppressing narrowband interferers.

RF watermarking is theoretically grounded in Dirty Paper Coding, which proves that a signal can be added to a channel already occupied by interference (the primary data) without reducing the primary channel's capacity, provided the watermark encoder has non-causal knowledge of that interference. In practice, the watermark modulator pre-distorts its tag using the known primary data symbols, effectively subtracting the primary signal's interference from the watermark at the moment of embedding. This allows the verifier to extract a clean tag.

The embedded tag is not a static identifier. It is a dynamic, cryptographically generated token that binds the device's secret key to a session-specific nonce or a hash of the current data payload. This prevents simple replay attacks where an adversary records and retransmits a valid watermarked signal. The verifier independently computes the expected tag for that session. A mismatch indicates either a spoofing attempt or a man-in-the-middle modification of the data payload, providing joint authentication of source and message integrity.

RF watermarking is an active, intentional authentication mechanism, distinct from passive RF fingerprinting which exploits unintentional hardware impairments. They are complementary layers of defense:

• RF Fingerprinting: Identifies a device's unique physical hardware signature.
• RF Watermarking: Validates a cryptographic claim of identity embedded in the signal. A robust physical layer security architecture combines both: the watermark provides a high-assurance cryptographic check, while the fingerprint detects hardware cloning or tampering that might compromise the key.

RF watermarking provides a continuous authentication channel independent of higher-layer protocols, making it ideal for zero-trust wireless networks. Use cases include:

• Critical Infrastructure: Verifying telemetry commands to power grid actuators without adding latency from cryptographic handshakes.
• Autonomous Vehicles: Continuously authenticating V2X safety messages to prevent ghost vehicle injection attacks.
• Military Communications: Providing transmission security (TRANSEC) by validating every burst at the physical layer before it reaches the network processor.

A hardware security primitive that exploits inherent manufacturing variations in silicon to generate a unique, unclonable identity. In RF watermarking, a PUF can be used to derive the secret key that modulates the covert authentication signal, ensuring the watermark is intrinsically bound to the specific device hardware and cannot be extracted or copied to another unit.

A technique that embeds the authentication tag by spreading its energy across a bandwidth much wider than the information signal. The watermark is modulated with a pseudo-noise (PN) sequence known only to the verifier, pushing its power spectral density below the noise floor. This makes the watermark virtually undetectable to an adversary without the spreading code, while the legitimate receiver can despread and recover it.

An information-theoretic coding scheme that allows a transmitter to embed a watermark into a primary signal without any interference between the two, provided the transmitter has non-causal knowledge of the interference (the primary signal itself). DPC achieves the theoretical maximum capacity for the watermark channel, making it the gold standard for transparent, high-rate embedding in known-host scenarios.

A structured sequence of physical-layer interactions designed to reliably verify device identity using the embedded watermark. A typical challenge-response protocol involves:

• The verifier sending a cryptographic nonce.
• The device embedding a response derived from the nonce and a secret key into its transmission.
• The verifier extracting and validating the response. This prevents replay attacks, as each authentication exchange is unique.

The core design goal of RF watermarking: the embedded authentication signal must be completely invisible to the intended data receiver. This is achieved by keeping the watermark power significantly below the primary signal's noise margin. A legacy receiver demodulates the data without any modification or performance degradation, while only an equipped verifier can extract the covert tag, enabling seamless overlay security.

The maximum rate at which authentication information can be reliably embedded and extracted without detection by an adversary. This capacity is fundamentally limited by the signal-to-noise ratio (SNR) of the primary channel and the acceptable watermark-to-signal power ratio. System designers must balance the need for robust, low-latency authentication against the risk of the watermark becoming observable or degrading the host communication.