Error Vector Magnitude

EVM aggregates I/Q imbalance, phase noise, amplifier non-linearity, and carrier leakage into a single scalar value. This aggregation creates a unique distortion signature because the specific combination and magnitude of each contributing impairment differs between individual transmitters due to manufacturing tolerances. The vector error at each symbol decision point captures the cumulative effect of all analog imperfections in the transmit chain.

The distribution of error vectors across the I/Q constellation reveals device-specific patterns. Different transmitters exhibit distinct error vector distributions even when achieving similar overall EVM values:

• Amplifier-dominated devices show larger errors at outer constellation points due to compression
• Phase-noise-dominated devices display rotational smearing proportional to symbol amplitude
• I/Q imbalance creates asymmetric error distributions between constellation quadrants These spatial error patterns provide discriminative features beyond the scalar EVM value.

EVM fingerprinting effectiveness varies with modulation density. Higher-order modulations like 256-QAM expose hardware impairments more prominently because the tighter decision boundaries amplify the visibility of small distortion vectors. A transmitter that appears nearly ideal under QPSK may reveal distinguishing error patterns under 64-QAM or 256-QAM. This sensitivity enables multi-modulation enrollment where devices are characterized across multiple constellation densities to build richer fingerprint profiles.

In OFDM systems, EVM measured per-subcarrier reveals frequency-selective impairment signatures. Filter ripple, impedance mismatches, and memory effects cause EVM to vary systematically across subcarriers. A transmitter's subcarrier EVM profile forms a distinctive pattern:

• Low-frequency roll-off indicates bias network time constants
• Periodic ripple reflects filter component tolerances
• Edge subcarrier degradation reveals amplifier memory effects This frequency-domain EVM signature provides high-dimensional fingerprinting data from a single transmission burst.

While individual symbol error vectors vary randomly due to thermal noise, the underlying EVM statistics remain stable over time for a given device. The mean EVM, error vector variance, and distribution shape constitute a persistent hardware signature. Long-term monitoring reveals slow drift patterns caused by temperature aging and component degradation, which themselves become identifying features. This temporal stability enables reliable re-identification of devices across multiple transmission sessions.

EVM-based fingerprinting requires channel equalization to isolate transmitter impairments from propagation effects. The measured EVM at the receiver includes contributions from multipath fading, Doppler shift, and additive noise. Advanced fingerprinting systems employ channel estimation and compensation before EVM calculation, using pilot symbols or blind equalization to remove channel-induced distortion. The residual EVM after equalization represents the transmitter-intrinsic impairment signature, enabling robust identification even in dynamic wireless environments.