Amplifier Non-Linearity

The deviation of a power amplifier's output amplitude from its ideal linear gain as input power increases. AM/AM distortion describes how the gain compresses near the 1 dB compression point (P1dB) and eventually saturates. This non-linear transfer function is unique to each physical amplifier due to semiconductor doping variances and transistor matching imperfections. The specific curvature of the gain compression curve serves as a robust, unclonable hardware fingerprint.

• Measured by comparing output power to input power across the dynamic range
• The third-order intercept point (IP3) quantifies the degree of non-linearity
• Subtle variations in the knee of the compression curve are device-specific

The unintended shift in the output signal's phase as a function of the instantaneous input amplitude. Unlike ideal linear amplifiers, real devices exhibit AM/PM conversion where stronger input signals cause greater phase lag due to the voltage-dependent capacitance of the transistor junctions. This phase distortion creates a unique signature in the phase trajectory of the transmitted signal, independent of the amplitude distortion, providing a second orthogonal dimension for device identification.

• Caused by non-linear parasitic capacitances in the transistor
• Measured in degrees per dB of input power change
• Creates asymmetric constellation warping in modulated signals

When a non-linear amplifier distorts a modulated signal, it generates intermodulation products that cause the signal's spectrum to spread into adjacent frequency channels. This spectral regrowth is a direct consequence of the amplifier's non-linear transfer function and manifests as an elevated noise floor in neighboring bands. The specific pattern and power of this leakage is highly device-dependent, as it is shaped by the unique AM/AM and AM/PM characteristics of each amplifier.

• Quantified by the Adjacent Channel Power Ratio (ACPR)
• The spectral shoulder shape is a distinctive fingerprint feature
• Becomes more pronounced as the amplifier approaches saturation

Power amplifier memory effects occur when the current output depends not only on the instantaneous input but also on previous signal states. This dynamic non-linearity is caused by thermal time constants (temperature changes in the transistor junction) and electrical time constants (bias circuit impedance variations with frequency). The resulting hysteresis in the AM/AM and AM/PM curves creates a signal-history-dependent signature that is exceptionally difficult to clone.

• Thermal memory: slow changes due to die heating over microseconds
• Electrical memory: fast changes from bias network impedance at envelope frequencies
• Creates asymmetrical distortion patterns in the constellation diagram

Non-linear amplification generates harmonic components at integer multiples of the fundamental carrier frequency. The relative power levels and phase relationships of the second, third, and higher-order harmonics are determined by the specific polynomial coefficients of the amplifier's transfer function. While harmonics are often filtered before transmission, their presence in near-field or unintentional emissions provides a rich, device-specific signature for radiated fingerprinting applications.

• Second harmonic (2f₀) is generated by even-order non-linearity
• Third harmonic (3f₀) is typically the strongest distortion product
• Harmonic phase coherence patterns are unique per device

The Volterra series provides a rigorous mathematical framework for modeling non-linear dynamic systems with memory. It represents the amplifier's output as a sum of multi-dimensional convolution integrals, capturing both instantaneous non-linearity and memory effects. The Volterra kernel coefficients extracted from a specific device form a compact, high-dimensional feature vector that uniquely characterizes its non-linear behavior for AI-based fingerprinting systems.

• First-order kernel: linear response
• Second-order kernel: quadratic non-linearity and memory
• Higher-order kernels capture progressively finer distortion details
• Truncated models (e.g., memory polynomial) are used for practical extraction

The fundamental characterization of amplifier non-linearity. AM/AM conversion describes the distortion of the output amplitude relative to the input amplitude, causing gain compression near saturation. AM/PM conversion describes the unwanted phase shift introduced as a function of instantaneous input power. Together, these two curves form a unique, device-specific mapping that serves as a robust fingerprint, as no two physical amplifiers exhibit identical conversion characteristics due to semiconductor manufacturing variances.

A dynamic non-linearity where the current output of a power amplifier depends not only on the present input but also on previous signal states. Caused by thermal time constants (self-heating of the transistor junction) and electrical time constants (bias network impedance and trapping effects in gallium nitride devices), memory effects create a distinctive signal-history-dependent signature. This temporal dependency adds a rich, multidimensional layer to the fingerprint that is extremely difficult to clone or replicate.

A complementary technique that applies an inverse model of the amplifier's non-linearity to the baseband signal before transmission. Neural network-based DPD systems learn to characterize the exact AM/AM and AM/PM curves of a specific amplifier to linearize its output. The residual distortion after pre-distortion—the uncorrectable error—is itself a highly unique fingerprint, as it represents the irreducible hardware-specific imperfections that no generic model can fully compensate.

A key figure of merit for amplifier linearity. The IP3 is a theoretical point where the power of the third-order intermodulation products would equal the desired fundamental signal power. A higher IP3 indicates better linearity. In fingerprinting, the specific IP3 value and the pattern of intermodulation distortion products generated when multiple tones are present provide a quantifiable, repeatable metric that is unique to each amplifier's semiconductor physics and biasing.

The output power level at which the amplifier's gain drops by 1 dB from its ideal linear value, marking the transition from linear to non-linear operation. The exact P1dB value, the shape of the gain compression curve approaching this point, and the asymmetry between the upper and lower compression characteristics are all device-specific. These subtle variations arise from doping inconsistencies and lithographic tolerances during semiconductor fabrication, making P1dB a stable, measurable fingerprint feature.

When a non-linear amplifier is driven by a modulated signal, intermodulation distortion causes the signal's spectrum to spread into adjacent frequency channels—a phenomenon called spectral regrowth. The specific pattern of this regrowth, measured as the Adjacent Channel Power Ratio (ACPR), is highly dependent on the amplifier's unique non-linear transfer function. The asymmetry and fine structure of the regrowth spectrum provide a rich, frequency-domain fingerprint that persists even through varying modulation schemes.