Power Amplifier Non-Linearity

The amplitude-to-amplitude (AM/AM) and amplitude-to-phase (AM/PM) transfer functions characterize how a PA compresses signal amplitude and introduces phase shifts as input power increases. These curves are unique to each amplifier due to process variation in transistor doping and gate oxide thickness.

• AM/AM compression: Gain reduction at high input power, causing constellation point warping
• AM/PM conversion: Phase rotation dependent on instantaneous envelope power
• Measured via swept-power vector network analyzer characterization
• Serves as a primary discriminant in Specific Emitter Identification (SEI) systems

When a PA is driven into compression, intermodulation distortion between spectral components causes energy to spill into adjacent frequency channels. This spectral regrowth pattern—its shape, asymmetry, and rate of decay—forms a distinctive signature.

• ACLR (Adjacent Channel Leakage Ratio) quantifies the power ratio between main and adjacent channels
• Asymmetry in upper vs. lower sideband regrowth indicates memory effects
• Third-order and fifth-order intermodulation products dominate the regrowth profile
• Highly sensitive to DC bias point and thermal state of the transistor

Non-linear amplification generates energy at integer multiples of the carrier frequency. The relative amplitudes and phases of these harmonic components (2nd, 3rd, 4th harmonics) are device-specific and can be captured for identification.

• Second harmonic (2f₀): Arises from asymmetric waveform clipping
• Third harmonic (3f₀): Dominant in differential PA topologies
• Harmonic phase relationships are more stable than amplitude over temperature
• Requires wideband receivers to capture out-of-band harmonic content
• Useful when in-band features are obscured by channel effects

PA memory effects cause the current output to depend on prior signal states due to thermal inertia, bias circuit time constants, and trapping effects in semiconductor materials. This creates a hysteresis-like distortion pattern unique to each device.

• Thermal memory: Die temperature changes with signal envelope, modulating gain on microsecond scales
• Electrical memory: Bias network impedance variations at envelope frequencies
• Trapping effects: Charge capture/release in GaN and LDMOS transistors
• Manifests as asymmetric intermodulation sidebands and dynamic AM/AM curve broadening
• Captured using two-tone or multi-tone test signals with varying spacing

The Volterra series provides a mathematical framework to model PA non-linearity with memory by representing the output as a sum of multi-dimensional convolution integrals. The extracted Volterra kernels serve as a compact, device-specific fingerprint.

• Linear kernel (1st order): Small-signal frequency response
• Nonlinear kernels (2nd, 3rd order): Capture intermodulation and harmonic generation
• Kernel asymmetry reveals memory effect characteristics
• Pruned Volterra models reduce coefficient count while preserving discriminative power
• Coefficients can be used directly as feature vectors for contrastive learning classifiers

Modern communication signals with high peak-to-average power ratio (PAPR) excite different regions of the PA transfer curve depending on instantaneous envelope amplitude. The resulting distortion pattern varies dynamically with signal statistics.

• Crest factor of the signal determines how often the PA enters compression
• Distortion is most severe at envelope peaks, creating transient signature bursts
• Complementary Cumulative Distribution Function (CCDF) of the signal interacts with PA non-linearity
• Different modulation formats (QPSK, 64-QAM, OFDM) stress the PA differently
• Enables cross-modulation fingerprinting where the same PA is identified across different waveforms

The two fundamental manifestations of power amplifier non-linearity. AM/AM distortion describes the non-linear relationship between input signal amplitude and output signal amplitude, causing gain compression at saturation. AM/PM distortion describes the unwanted phase shift introduced as a function of input amplitude. The specific curves of these distortions form a core part of a device's RF-DNA.

The phenomenon where non-linear amplification causes a modulated signal's bandwidth to spread into adjacent frequency channels, creating Adjacent Channel Power (ACP). The specific pattern and level of this regrowth is a direct consequence of the amplifier's non-linear characteristics and is heavily regulated. For fingerprinting, the unique spectral regrowth shape serves as a robust, frequency-domain feature for emitter identification.

A critical complexity in power amplifier modeling where the current output depends not only on the instantaneous input but also on past inputs. These effects, caused by thermal dynamics, biasing circuit time constants, and charge trapping, create a unique, time-dependent distortion signature. Capturing these effects is essential for high-fidelity behavioral models like the Volterra series used in advanced fingerprinting.

A key figure of merit for quantifying a power amplifier's non-linearity. The IP3 is a theoretical point where the power of the desired fundamental signal and the undesired third-order intermodulation products would be equal. A higher IP3 indicates better linearity. This metric is used to benchmark amplifiers, and the specific, measured IP3 value is a hardware-intrinsic characteristic.

A powerful mathematical framework for modeling non-linear dynamic systems with memory, including power amplifiers. The Volterra series expresses the output as a sum of multi-dimensional convolution integrals, capturing both harmonic distortion and intermodulation products. The specific Volterra kernels extracted from a device's signal provide a comprehensive, unique fingerprint for robust identification.