Envelope Tracking¶
State estimation — instantiates Wave Packet Propagation and Spreading
Measures the packet's whole evolving shape — center, width, edges, and concentration — so decisions rest on the full envelope instead of a single point arrival.
Envelope Tracking is the shape estimator of the archetype: it reconstructs and maintains the packet as a finite, evolving distribution — a center, a width, a leading and trailing edge, an amplitude or concentration, and a support region — from whatever measurements are coming in. Its defining move is that it reports the currently measured shape, not a forecast of where the packet is headed and not an action to change it. Where a forward model predicts and an intervention steers, Envelope Tracking simply refuses to collapse the packet to a point: instead of "42 cases this week" or "the front has arrived," it produces a moving picture of the whole envelope. That picture is the raw material every other mechanism consumes, and its whole value is that it keeps the center, front, tail, and width from being silently conflated.
Example¶
A regional health department is following a seasonal influenza wave. The reflex is to report one number — weekly confirmed cases — and call the peak when that number turns down. Envelope Tracking replaces the number with a shape. Fusing sentinel-clinic visits, hospital admissions, and a wastewater viral signal, it reconstructs the wave as a distribution: a leading edge already climbing in the northern counties, a center of mass drifting south at roughly a county a week, a peak amplitude, and a long convalescent tail of lingering hospitalizations that persists well after case counts fall. The reconstructed envelope tells planners three things a single count hid: the southern districts have not yet crested, the tail will keep ICU occupancy elevated for weeks after the "peak" is declared, and the northern and southern signals are two parts of one moving packet rather than two separate outbreaks. Staffing follows the envelope, not the headline count.
How it works¶
The method fuses heterogeneous, differently-lagged sensors into one reconstructed profile, then reads its geometry: the center as a first moment, the width as a spread (second moment), the leading and trailing edges as chosen percentiles of the support, and the amplitude as peak concentration or density. It then propagates the estimate through time by tracking how each of those quantities moves — the center advancing, the width broadening, the tail lengthening. Crucially it interpolates the measured shape rather than solving a transport equation; it is a filter over observations, not a physics prediction, which is exactly what keeps it distinct from the forecasting mechanisms it feeds.
Tuning parameters¶
- Moment order — track just center-and-width, or add higher moments for skew and tail heaviness. Richer moments catch an asymmetric tail but demand more clean data.
- Smoothing window — how much temporal smoothing to apply. More smoothing steadies a noisy estimate but lags fast changes and can blur a genuine narrowing into a plateau.
- Edge definition — which percentile or threshold marks the leading and trailing edges. Tight edges follow the core; loose edges include faint but real tails.
- Sensor-fusion weights — how much to trust each heterogeneous source. Over-weighting a biased sensor tilts the entire reconstructed envelope.
- Amplitude normalization — absolute concentration versus shape normalized to unit area. Normalizing lets you compare shape across time but hides changes in raw intensity.
When it helps, and when it misleads¶
Its strength is that it directly cures the archetype's two headline errors — point-collapse and front-envelope confusion — by making the tail, the asymmetry, and the total footprint visible instead of inferred. It gives every downstream mechanism a shared, honest picture of what the packet actually looks like right now.
Its central failure mode is that, in real time, the trailing edge is right-censored[n1]: you only ever see the part of the packet that has already been observed, so the tail length and total footprint are systematically under-called until later data backfills them. The classic misuse is to read the tracked peak as the peak while the wave is still rising, freezing a staffing or containment decision on a shape that isn't finished. A subtler trap is to accept a smoothing-blurred single bump as ground truth when it is really two merging branches — a question Envelope Tracking cannot answer and must hand off. The guarding discipline is to carry an explicit real-time-versus-revised band on the tail, and to route any suspiciously wide or double-shouldered envelope to a dedicated splitting check rather than trusting the smooth reconstruction.
How it implements the components¶
Envelope Tracking fills the representation slots of the archetype — the shape of the packet, not its future or its control:
localized_packet_profile— it reconstructs and maintains the profile itself: center, width, envelope, amplitude/concentration, and support region, kept current as measurements arrive.packet_motion_estimate— it tracks how the center, leading edge, trailing edge, and support region move over time, keeping arrival timing distinct from broadening.
It does not design where or when to sample (observation_window_and_sampling_plan) or run any forward forecast (ensemble_forecast_grid, spreading_and_dispersion_model) — those belong to its nearest twin, Adaptive Resampling and Reforecasting; Envelope Tracking reports the shape it has measured, never the shape it predicts. It also follows the amplitude envelope, not the phase beneath it, so carrier_phase_or_internal_state_tracking is Packet Splitting and Recombination Detection's.
Related¶
- Instantiates: Wave Packet Propagation and Spreading — Envelope Tracking supplies the measured packet shape the rest of the archetype reasons over.
- Sibling mechanisms: Adaptive Resampling and Reforecasting · Attenuation, Damping, and Absorption · Boundary Reflection, Absorption, or Channeling · Dispersion Compensation or Refocusing · Packet Splitting and Recombination Detection · Advection-Diffusion or Transport Modeling
Editorial Notes¶
Form Classification¶
Form family: Monitoring, Sensing & Alerting
Rationale: The mechanism repeatedly fuses actual sensor observations into a current packet profile and tracks its center, width, edges, concentration, and evolution through time.
Nearest alternative: Analysis, Modeling & Optimization — Moment and percentile calculations estimate state, but the mechanism is explicitly a continuing observation filter rather than a physics prediction or offline model.
Review outcome: Adjudicated after independent review; high confidence.
Origin Attribution¶
Primary origin: Physics
Origin pattern: Single lineage
Present-day reach: Specialized
Rationale: Wave and signal physics cohered envelope descriptions that track a packet's center, width, edges, and spreading rather than a single point.
Related originating lineages:
- Engineering & Design — Signal-processing and communications engineering operationalized envelope detection and tracking.
Review resolution: The current reviewers agree that physics is primary. For the reported differences (alternate_origin_disagreement), the evidence supports single_lineage, specialized, and engineering_design; these choices preserve materially formative origins without conflating later domain reach.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] Right-censoring is the statistical condition of observing only events that have happened by the present, with later ones still pending. Applied to a moving packet, it means the trailing edge and total footprint are only ever partially observed in real time — the reason a live envelope estimate must carry a revision band on its tail. ↩