I/Q Imbalance

Gain imbalance occurs when the I-branch amplifier and Q-branch amplifier have mismatched gain factors, causing one component of the signal to be scaled differently than the other.

• Mathematical representation: The transmitted signal becomes s(t) = I(t) * G_I * cos(ωt) + Q(t) * G_Q * sin(ωt) where G_I ≠ G_Q
• Constellation effect: A perfect square QPSK constellation stretches into a rectangular shape
• Typical values: Even 0.5-2% gain mismatch creates measurable, device-specific distortion
• Stability: Gain ratios remain remarkably stable over temperature, making them excellent long-term fingerprints

Phase imbalance, also called quadrature error, occurs when the I and Q local oscillator signals are not exactly 90 degrees apart, introducing cross-talk between the two signal components.

• Origin: Imperfect 90-degree phase shifters in the quadrature modulator
• Constellation effect: The constellation rotates and skews, with points appearing as a parallelogram rather than a square
• Measurement: Expressed in degrees of deviation from ideal quadrature (e.g., 2° phase error)
• Fingerprinting value: Phase imbalance patterns are unique to each transmitter's mixer and LO path layout

I/Q imbalance is categorized by its behavior across the signal bandwidth, which determines the complexity of both compensation and fingerprint extraction.

• Frequency-independent imbalance: Constant gain and phase error across all subcarriers, caused by mixer and baseband amplifier mismatches
• Frequency-dependent imbalance: Varying error across frequency, caused by I/Q filter mismatches in the analog baseband chain
• OFDM impact: In multi-carrier systems, frequency-dependent imbalance creates mirror-subcarrier interference where subcarrier k leaks into subcarrier -k
• Fingerprint richness: Frequency-dependent imbalance provides a multidimensional signature with higher discriminative power

The Image Rejection Ratio (IRR) quantifies the severity of I/Q imbalance by measuring the power ratio between the desired signal and its unwanted image.

• Definition: IRR = 10 * log10(P_desired / P_image) expressed in dB
• Typical values: Uncompensated consumer devices exhibit 25-40 dB IRR; high-precision equipment achieves 50-60+ dB
• Fingerprinting application: The IRR value and its variation across frequency, power levels, and temperature form a multi-parametric device signature
• Measurement technique: Transmit a single-tone test signal and measure the power of the resulting image tone at the negative frequency

I/Q imbalance rarely occurs in isolation; its interaction with local oscillator leakage creates compound distortions that enhance fingerprint uniqueness.

• Combined effect: DC offset shifts the constellation origin, while I/Q imbalance skews and stretches it, producing a unique asymmetric constellation
• Modeling: The combined impairment is modeled as y(t) = α * x(t) + β * x*(t) + d where α captures gain/phase imbalance, β captures image leakage, and d is the DC offset vector
• Discriminative power: The joint parameter space {α, β, d} provides a high-dimensional fingerprint that is extremely difficult to clone or emulate
• Extraction: Blind estimation algorithms can separate these parameters from received signals without known training sequences

The long-term stability of I/Q imbalance under environmental variation is critical for reliable device authentication over months or years of operation.

• Temperature sensitivity: Gain imbalance drifts slowly with temperature (typically 0.01-0.05% per °C) due to amplifier biasing changes
• Phase stability: Quadrature error is relatively temperature-insensitive as it depends primarily on passive component matching in the phase shifter
• Aging effects: Component aging over years causes monotonic drift that can be tracked with adaptive baseline updating algorithms
• Compensation: Drift compensation models predict and adjust for these slow variations, maintaining authentication accuracy without requiring frequent re-enrollment

The direct visual manifestation of I/Q imbalance on a signal's constellation diagram. Gain mismatch causes the ideal square constellation to become rectangular, while phase error skews the axes, warping the diagram into a parallelogram. This static, geometric distortion is a primary feature for fingerprinting because it is a persistent, hardware-specific signature that can be extracted from the steady-state portion of any digitally modulated transmission.

A closely related impairment often co-occurring with I/Q imbalance. It results from a portion of the unmodulated carrier signal leaking through the mixer, creating a DC offset in the baseband constellation. This shifts the entire constellation away from the origin. Together, I/Q imbalance and LO leakage define the fundamental linear distortions of a transmitter's quadrature modulator, forming a composite, unclonable hardware signature.

A comprehensive, aggregate metric that quantifies the deviation of measured constellation points from their ideal reference positions. I/Q imbalance is a major contributor to EVM degradation. While EVM provides a single quality score, fingerprinting systems decompose it to isolate the specific gain and phase error components, using these individual impairment parameters as discriminative features rather than the combined magnitude.

Represents the primary non-linear impairment in a transmitter chain, in contrast to the linear distortion of I/Q imbalance. PA non-linearity causes AM-AM and AM-PM conversion, warping the outer constellation points more than the inner ones. Advanced fingerprinting systems must disentangle these non-linear effects from the linear I/Q imbalance to create a robust, power-independent device signature.

A technique where a neural network learns to apply an inverse distortion to a signal before the power amplifier, linearizing the output. While designed to correct impairments, the pre-distortion coefficients themselves become a unique signature. A device's specific correction vector, learned to counteract its unique I/Q imbalance and PA non-linearity, serves as a highly stable, machine-readable fingerprint.

The mathematical process of transforming raw I/Q samples into a compact, numerical representation for a classifier. For I/Q imbalance, this involves estimating the gain mismatch (α) and phase error (θ) parameters directly from the received signal using blind estimation algorithms. These two scalar values, or the elements of the resulting correction matrix, form a low-dimensional but highly discriminative feature vector for device identification.