Constellation Cloud

The constellation cloud represents the probability density function of received symbols around an ideal locus. Unlike deterministic I/Q imbalance, this dispersion is stochastic, driven by additive white Gaussian noise (AWGN), phase noise, and inter-symbol interference (ISI). The cloud's morphology—its spread, symmetry, and kurtosis—forms a unique noise signature. Key geometric properties include:

• Centroid offset: The mean displacement from the ideal point, indicating static DC offset.
• Cloud ellipticity: The ratio of major to minor axis spread, revealing correlated noise sources.
• Angular dispersion: The variance in phase, directly quantifying local oscillator phase noise.

Phase noise is a primary contributor to the constellation cloud's angular spread. It originates from short-term instabilities in the local oscillator (LO), causing random phase modulation of the carrier. In the constellation diagram, this manifests as a tangential smearing of points along an arc centered at the origin. The power spectral density of this phase noise, often modeled with a Leeson's equation profile, creates a distinct signature. Key characteristics:

• Close-in phase noise causes slow, correlated rotation of the entire constellation.
• Far-out phase noise produces independent, symbol-to-symbol angular jitter.
• The variance of the angular error is a robust feature for emitter identification.

Additive white Gaussian noise (AWGN) creates a radially symmetric, Gaussian distribution around each ideal constellation point. The signal-to-noise ratio (SNR) directly dictates the cloud's radius: a lower SNR produces a larger, more diffuse cloud. However, the noise is not always perfectly white; colored noise from power supplies or digital switching can create anisotropic cloud shapes. Analysis techniques include:

• Error Vector Magnitude (EVM) provides a single scalar measure of the cloud's RMS size.
• Per-symbol variance analysis reveals if noise power is signal-level dependent.
• The shape of the noise cloud, beyond its size, can indicate specific non-ideal amplifier behaviors.

ISI, caused by channel multipath or inadequate pulse shaping, smears the energy of one symbol into adjacent symbol periods. In the constellation cloud, this creates a deterministic, data-dependent dispersion that is distinct from random noise. The cloud points form specific trajectories between ideal loci based on the transmitted symbol sequence. Key indicators of ISI:

• Cloud elongation along a line connecting adjacent constellation points.
• Eye diagram closure, which directly correlates with the cloud's vertical spread.
• Unlike noise, ISI-induced dispersion is predictable and can be used to characterize the channel or transmitter filter imperfections.

The internal structure of a constellation cloud is a rich source of fingerprinting features. Gaussian Mixture Models (GMMs) and DBSCAN are used to analyze the density distribution of measured points. A device with a clean power supply might show a tight, single-mode Gaussian cloud, while one with a faulty amplifier might exhibit a multi-modal or skewed distribution. Extracted features include:

• Kurtosis: A measure of the cloud's tailedness, indicating impulsive noise.
• Cluster cardinality: The number of distinct sub-clusters within a symbol's cloud.
• Divergence metrics like KL-divergence quantify how a device's cloud differs from an ideal Gaussian reference.

A constellation cloud is not static; its characteristics evolve over time due to thermal drift and component aging. Monitoring the slow migration of the cloud's centroid and the change in its dispersion provides a dynamic signature. Key temporal behaviors:

• Warm-up drift: The rapid change in cloud parameters as a device reaches thermal equilibrium after power-on.
• Long-term aging: A gradual, monotonic shift in the cloud's shape over months or years.
• Environmental sensitivity: The correlation between cloud metrics and ambient temperature, which can be modeled and compensated for to maintain authentication accuracy.

The primary comprehensive metric for quantifying constellation cloud dispersion. EVM measures the magnitude of the error vector between the measured symbol location and its ideal reference point.

• Expressed as a percentage of the reference signal amplitude
• Directly correlates with signal-to-noise ratio (SNR)
• A larger EVM indicates a wider constellation cloud and poorer modulation accuracy
• Used as a pass/fail metric in wireless standards like IEEE 802.11 and 3GPP

A two-dimensional scatter plot that visualizes the constellation cloud by plotting the in-phase (I) component on the x-axis against the quadrature (Q) component on the y-axis. Each cluster of points represents a symbol.

• The constellation cloud appears as a spread of points around each ideal symbol locus
• Systematic distortions like I/Q imbalance warp the cloud into elliptical shapes
• DC offset displaces the entire cloud from the origin
• Provides immediate visual feedback on transmitter hardware health

A figure of merit representing the average power ratio of the ideal reference signal to the error vector power within the constellation cloud. MER is the logarithmic counterpart to EVM.

• Calculated as 10 * log10(Psignal / Perror) in dB
• A higher MER indicates a tighter constellation cloud and cleaner signal
• Commonly used in digital video broadcasting (DVB) and cable modem (DOCSIS) standards
• Provides a single-number summary of modulation fidelity

A primary physical mechanism that broadens the constellation cloud through random phase modulation of the carrier. Phase noise originates from instabilities in the local oscillator and manifests as angular smearing.

• Causes the constellation cloud to spread tangentially around the origin
• Higher-order modulations like 1024-QAM are more susceptible to phase noise
• Characterized by its single-sideband phase noise spectrum (dBc/Hz)
• Different oscillator technologies (TCXO, OCXO) produce distinct noise signatures

The fundamental thermal noise present in all electronic systems that creates a perfectly circular, symmetric constellation cloud around each ideal symbol point. AWGN sets the theoretical Shannon capacity limit.

• Produces a Gaussian distribution of error vectors in both I and Q dimensions
• The constellation cloud radius is proportional to the noise power spectral density (N0)
• Unlike hardware impairments, AWGN is not a unique identifier
• SNR directly determines the cloud size relative to symbol spacing

A deterministic distortion that elongates the constellation cloud when energy from adjacent symbols spills into the current symbol period. ISI is caused by multipath propagation or bandlimiting filters.

• Creates a smeared, non-Gaussian constellation cloud shape
• Equalization techniques like decision feedback equalizers (DFE) can mitigate ISI
• The resulting cloud pattern depends on the specific data sequence transmitted
• Severe ISI closes the eye diagram and increases bit error rate (BER)