Intermodulation Distortion (IMD)

IMD products directly map to the coefficients of a device's power series model. By measuring the amplitude and phase of 3rd-order (2f1-f2, 2f2-f1) and 5th-order products, the specific static non-linearity polynomial can be reverse-engineered. This polynomial serves as a compact, highly discriminative fingerprint vector for emitter identification.

The standard method for characterizing IMD uses two closely spaced, equal-amplitude continuous wave (CW) tones. The frequency spacing is critical:

• Narrow spacing (e.g., 100 kHz) reveals memory effects in the power amplifier, as thermal and electrical time constants modulate the distortion.
• Wide spacing (e.g., 10 MHz) isolates static non-linearity, as the device cannot thermally react to the beat frequency. This dual approach separates static and dynamic fingerprint components.

The Input-referred Third-Order Intercept Point (IIP3) is a figure of merit extrapolated from IMD measurements. It represents the theoretical input power where the fundamental and 3rd-order IMD products would intersect. IIP3 is not a constant but a device-specific, frequency-dependent parameter that varies due to process-voltage-temperature (PVT) variations, making it a robust, scalar fingerprint feature.

When stimulated with modulated signals (e.g., OFDM), IMD manifests as spectral regrowth into adjacent channels. This is quantified by the Adjacent Channel Leakage Ratio (ACLR). The specific shape and asymmetry of this regrowth spectrum are highly sensitive to the AM-AM and AM-PM distortion characteristics of the power amplifier, creating a complex, frequency-domain fingerprint that is difficult to clone.

A purely memoryless non-linearity produces symmetric upper and lower IMD sidebands. Asymmetry in the amplitude or phase of these sidebands is a direct indicator of electrical (bias network) and thermal memory effects. This asymmetry is a complex, time-dependent fingerprint feature that captures the dynamic interaction between the transistor, its package, and the power supply, providing a signature that is extremely challenging to replicate with a simple static model.

While a Taylor series models static non-linearity, a Volterra series captures frequency-dependent dynamic non-linearity by incorporating memory kernels. The Volterra kernels extracted from multi-tone IMD measurements provide a complete, predictive model of a device's non-linear behavior, including frequency-dependent gain compression and desensitization. This full kernel set constitutes a high-dimensional, unforgeable fingerprint.

A key figure of merit for quantifying third-order non-linearity. IP3 is a theoretical point where the power of the third-order intermodulation products would equal the power of the fundamental tones. A higher IP3 indicates better linearity. It is a critical parameter for predicting a device's non-linear signature and modeling the generation of IMD products like 2f1 - f2 and 2f2 - f1.

The ratio of the sum of the powers of all harmonic components to the power of the fundamental frequency. While IMD involves multiple input signals, THD quantifies the distortion a single tone experiences. The same non-linear transfer function that creates harmonics is responsible for intermodulation products, making THD a directly related measure of a device's intrinsic non-linear fingerprint.

The ratio of the RMS amplitude of the fundamental signal to the highest spurious component in the output spectrum. This 'spur' is often an intermodulation product. SFDR is a critical specification for identifying device-specific non-linear artifacts, as the frequency and magnitude of the largest IMD product directly reveal the polynomial coefficients of the transmitter chain's transfer function.

A memoryless, amplitude-dependent distortion modeled by a polynomial transfer function. This is the direct mathematical cause of Intermodulation Distortion. The polynomial coefficients (a2, a3, etc.) are unique to each device due to manufacturing variances. Analyzing the resulting IMD products allows for the reverse-engineering of these coefficients, forming a consistent, time-invariant component of the RF fingerprint.

A dynamic non-linearity where the current output depends on past signal values, often due to thermal or electrical time constants. This causes the IMD products to become asymmetric and frequency-dependent, creating a much richer and more complex signature than static non-linearity alone. Volterra series are often used to model this behavior, making the fingerprint significantly harder to clone.

A technique that intentionally pre-distorts a signal with the inverse of a power amplifier's non-linear transfer function to linearize the output. While DPD aims to suppress IMD, the residual distortion and the specific DPD correction coefficients themselves become a unique, learnable feature of the transmitter. The act of compensating for IMD creates a new, complex fingerprint.