Burst Offset Detection

The core function is to locate the exact sample index where the signal burst ends. This requires distinguishing the decaying ramp-down signature from the background noise floor with microsecond or nanosecond precision. Algorithms must avoid false triggers on transient spectral splatter or ringing artifacts that extend beyond the main envelope collapse.

Static amplitude thresholds fail in dynamic electromagnetic environments. Robust detection employs adaptive thresholding based on real-time noise floor estimation:

• Constant False Alarm Rate (CFAR) algorithms dynamically adjust the detection threshold to maintain a constant probability of false alarm.
• Noise floor tracking uses a sliding window or recursive estimator to follow slow variations in background interference.
• Hysteresis prevents chattering at the boundary by using separate thresholds for the start and end of the offset event.

Rather than operating on raw oscillating carrier samples, detection is often performed on the signal envelope extracted via the Hilbert transform. The envelope's transient decay profile provides a smoother, unipolar signal for analysis. Key metrics include:

• Fall-time variance: Statistical analysis of the 90% to 10% amplitude collapse duration.
• Burst trailing edge slope: The maximum negative rate of change, calculated as the first derivative of the envelope.
• Energy envelope collapse: Monitoring the squared magnitude to detect the moment energy transfer ceases.

The turn-off transient often includes an abrupt, unintended phase discontinuity as the frequency synthesis components power down. Detection algorithms can exploit this by monitoring the instantaneous phase trajectory for sudden, non-linear jumps. Zero-crossing analysis of the raw IQ samples can reveal timing anomalies that mark the precise offset moment, independent of amplitude fluctuations that may confuse envelope-only detectors.

The transition from a structured signal burst to stochastic noise is a statistical change-point problem. Algorithms like the Bayesian Change Point Detector or Cumulative Sum (CUSUM) test monitor the likelihood function of the incoming samples. A significant shift in the statistical distribution—such as a drop in transient kurtosis or a change in spectral flatness—indicates the burst offset boundary with high mathematical rigor.

A primary challenge is trailing edge jitter, the timing variation in the falling edge across multiple bursts from the same device. This jitter, caused by power supply decoupling inconsistencies and logic gate propagation delays, can smear the detected offset point. Robust detection systems must characterize this jitter statistically and align multiple captures using transient correlation fingerprinting techniques to isolate the consistent underlying hardware signature from stochastic timing noise.

The complementary signal processing algorithm that precisely locates the temporal boundary where an RF transmission transitions from the noise floor to an active state. While burst offset detection identifies the end of a burst, onset detection captures the leading edge.

• Uses adaptive thresholding against the noise floor
• Often employs a dual-threshold hysteresis to prevent false triggers from noise spikes
• Critical for isolating the turn-on transient for fingerprint extraction
• Typically implemented via short-time energy detectors or wavelet-based changepoint detection

The short-duration signal anomaly generated during the power-down sequence of a transmitter. This transient is characterized by unique phase discontinuities and amplitude collapse profiles that reflect the discharge behavior of capacitive elements.

• Reveals the power supply holdup capacitance and discharge path impedances
• Contains a distinct non-linear amplitude decay unlike the ramp-up profile
• Often exhibits transient phase noise bursts as the PLL loses lock
• The fall-time variance across multiple bursts provides a statistical device identifier

The extraction of the instantaneous magnitude contour of a transient signal, typically computed using the Hilbert transform. This isolates the amplitude modulation envelope from the carrier oscillations.

• Characterizes the attack, decay, sustain, and release (ADSR) profile of a burst
• The Hilbert Transform Envelope provides a precise amplitude contour without carrier-cycle distortion
• Enables measurement of rise-time variance and fall-time variance
• Forms the basis for extracting overshoot, undershoot, and ringing artifact features

The measurement of the duration required for a transmitter's frequency and amplitude to stabilize within a specified tolerance after the initial turn-on event. This reveals the dynamic characteristics of the phase-locked loop (PLL) and bias circuitry.

• PLL lock time is a critical component, exposing loop filter dynamics
• Frequency settling profile traces the carrier's convergence trajectory
• Component tolerances in the loop filter create unique settling signatures
• Often analyzed via transient frequency trajectory plots in the time-frequency domain

A damped sinusoidal oscillation superimposed on the transient envelope, caused by parasitic inductance and capacitance resonating in the transmitter's output matching network. The damped oscillation profile serves as a distinct hardware signature.

• Characterized by its resonant frequency and exponential decay constant (time constant)
• Caused by the reactive components in the output matching network
• The Q-factor of the ringing is determined by the effective resistance in the resonant tank
• Highly repeatable within a device but varies between units due to manufacturing tolerances

An abrupt, unintended shift in the instantaneous phase of a carrier signal during the turn-on or turn-off transient. This is caused by the non-ideal switching of frequency synthesis components and modulator settling.

• Measured via transient phase trajectory in the complex IQ plane
• Results from transient DC offset causing carrier feedthrough
• The magnitude and direction of the phase jump are device-specific
• Can be analyzed using transient differential constellation plots of successive IQ samples