19 WSN Tracking Fundamentals

wireless-sensor-networks

target-tracking

fundamentals

19.1 Start With the Field Story

This section: Start With the Field Story — Picture a lone worker moving through a large plant during an inspection.

Carry One Track Across the Handoff

Picture a lone worker moving through a large plant during an inspection. One group of sensors sees the badge near a pump, then another group sees a weak trace near the next room. The safety desk must decide whether both observations belong to the same person and whether the track is still current.

Cartoon panel 1 illustrates this step: Two sensor groups track one worker between factory rooms. — A worker's track keeps its evidence through a sensor-group handoff.

Cartoon panel 2 illustrates this step: A weak or late handoff can join the wrong sightings. — A worker's track keeps its evidence through a sensor-group handoff.

Build one track record from direct observations. Keep the target label, event time, source group, estimated place, doubt, prediction age, and current owner. Mark when the record changes from seen to predicted, lost, or found again. Never present an estimate as a direct sighting.

Test a missed reading, two nearby targets, a late handoff, a sleeping relay, a failed node, and a clock shift. Check what the next group received and what it confirmed. A line on a map is not proof that identity and time survived the handoff.

Keep urgent worker safety based on the freshest local evidence and a clear lost state. A remote history can aid review, but it must not turn an old estimate into a live location.

This opening does not select a sensing method or promise exact position. Practitioner builds the handoff and release rules. Under the Hood examines location clues, fusion, energy, prediction, doubt, and recovery across a changing field.

Tracking begins when one reading is not enough. A WSN must connect observations over time, manage uncertainty, hand off between nodes, and decide when an estimate is good enough for action without pretending that every target path is known.

Wireless sensor network target tracking is the work of keeping a moving target record honest as evidence moves from sensing nodes to an operator, gateway, or automated decision. A tracking page should not only say that a target was found. It should show what was observed, what was estimated, which sensors owned the record, when the state changed, and when uncertainty became too large to trust.

The core difference between detection and tracking is continuity. Detection asks whether a target was observed. Tracking asks whether the network can maintain a labeled target state as the target moves across coverage areas, gateways, failure modes, and time.

19.2 In 60 Seconds

This section: In 60 Seconds — A tracking record should separate observation, estimate, prediction, and release label.

A tracking record should separate observation, estimate, prediction, and release label. Handoffs are the main continuity risk: the next sensor group must know what evidence it received, what it confirmed, and what remains uncertain. Localization evidence can come from signal strength, timing, angle, proximity, camera or acoustic detections, and fusion, but every position estimate needs a quality label. Energy efficiency comes from selective activity, not from hiding missed observations. Sleeping, waking, sampling, relaying, and recovery all belong in the evidence record. A tracked target should be labeled as current, predicted, delayed, ambiguous, lost, recovered, partial, or historical before a learner or operator acts on it.

19.3 Learning Objectives

This section: Learning Objectives

By the end of this chapter, you will be able to:

Explain the difference between detection, localization, prediction, and tracking continuity.
Identify the evidence a WSN target record must preserve before it is released.
Review handoff and recovery points where a moving target can be mislabeled.
Describe how energy state, measurement quality, and gateway custody affect tracking claims.
Apply release labels before approving a tracking fundamentals page, lab, simulation, or prototype.

19.4 WSN Tracking Fundamentals

This section: WSN Tracking Fundamentals

19.5 Tracking Claim

This section: Tracking Claim — This claim avoids three shortcuts.

Start with a claim that can be reviewed:

Tracking fundamentals review claim: The system can maintain a target state over time by preserving observation evidence, position estimates, uncertainty, sensor ownership, energy state, gateway custody, handoff history, recovery actions, and release labels.

This claim avoids three shortcuts. It does not treat one detection as a track. It does not treat a prediction as an observation. It does not treat an energy-saving sleep schedule as acceptable unless the page also shows how missed, stale, or recovered target states are handled.

Observation Which nodes sensed the target, when, with what measurement type and quality?

Estimate How was the target position or zone inferred from observations and geometry?

Continuity Which node group owns the track now, and what happens at the next handoff?

Release label Is the target current, predicted, ambiguous, lost, recovered, delayed, partial, or historical?

19.6 Tracking Evidence Loop

This section: Tracking Evidence Loop — Before applying Tracking Evidence Loop, view Figure 19.1 and compare INPUT with Observe.

Before applying Tracking Evidence Loop, view Figure 19.1 and compare INPUT with Observe. Their relationship makes the evidence behind Tracking Evidence Loop reviewable.

WSN tracking fundamentals evidence loop showing observation, localization, state estimate, handoff ownership, recovery, gateway custody, and release labels. — Figure 19.1: WSN tracking fundamentals evidence loop separating observation, localization, estimate, handoff, recovery, and release labels.

Figure Figure 19.1 follows a worker’s badge observation between sensor groups. The first group sees the badge near a pump and keeps the event time and source. Handoff must pass the last observation, prediction boundary, confidence label, and owner action if the worker does not appear. A weak trace near the next room may confirm, correct, or leave the track uncertain. The loop below separates observation, localization, estimate, handoff, release, and recovery. This prevents a late or weak handoff from joining unrelated sightings or showing a predicted state as a fresh direct observation.

The evidence loop starts with a sensor observation and ends with a released state. The loop repeats whenever the target moves, a sensor sleeps, a gateway receives a summary, or the track is recovered after a gap.

The next Tracking Evidence Loop step depends on the evidence behind Tracking Evidence Loop. Read Figure 19.2 first, focusing on WSN Tracking Interactive Evidence Record and SCENARIO.

Worker track crosses from one sensor group to another near a pump room. The record marks an observed trail at time t-zero, a predicted state for t-one with confidence bounds, and a weak new trace that must be compared before a confirmed label or action is released. — Figure 19.2: Worker-tracking handoff record separates the last confirmed sensor observation from a predicted next position and a weak new trace, retaining event time, node identity, quality, and confidence through the handoff.

Figure Figure 19.2 makes the handoff record explicit. It keeps a confirmed observation at time t₀ apart from a predicted state for t₁ and a weak new trace at t₁. The observation field needs node, mode, time, quality, and whether evidence is direct. Localization adds method and confidence boundary; the estimate adds age, assumptions, and retirement rule. Release records custody, safe conclusion, state label, and limits. A line across the map cannot prove identity or timing. The next sensor group must confirm continuity before the dashboard can call the target current.

Observation record Record node id, sensing mode, timestamp, signal quality, local noise, and whether the observation was direct or inferred.

Localization record State the method used: proximity, range estimate, angle estimate, time difference, visual cue, acoustic cue, or fused evidence.

State estimate Keep position, zone, velocity, confidence boundary, age, and the assumptions used to create the estimate.

Release record Show what the learner or operator can safely conclude and what remains unproven.

19.6.1 Detect, Localize, Predict, Report

A tracking cycle is easier to review when the four actions are separated:

Detect A node senses the target or event at a stated time with a stated measurement quality.

Localize The system turns one or more observations into a zone, position, path segment, or confidence boundary.

Predict The system estimates the next possible state so the right sensors, relays, or gateways can prepare.

Report The released state carries source evidence, age, uncertainty, custody, and a label the user can trust.

The loop does not make prediction the same as observation. A current label needs recent evidence. A predicted label needs a last observation, a bounded motion assumption, and a retirement rule.

19.7 Knowledge Check: Tracking Cycle

This section: Knowledge Check: Tracking Cycle

19.8 Detection Is Not Tracking

This section: Detection Is Not Tracking — A single detection can be useful, but it is not a track by itself.

A single detection can be useful, but it is not a track by itself. Tracking requires a sequence of states that remain interpretable as the target moves through the network.

Detection question Was there evidence of a target near a sensor, zone, or boundary at a stated time?

Tracking question Can the network preserve a target state across time, movement, sensor handoff, and missing evidence?

Localization question What position, zone, or path is supported by the available measurement geometry?

Prediction question What possible future state follows from the last evidence, and when must that prediction retire?

A common quality problem is a page that moves from “a node detected motion” to “the target is at this location” without explaining geometry, uncertainty, or handoff. The review should force that missing step into the open.

19.9 Measurement Evidence

This section: Measurement Evidence — Wireless tracking can use several measurement families.

Wireless tracking can use several measurement families. Each family has different uncertainty, calibration, and failure behavior. The chapter should explain the evidence boundary instead of promising a universal accuracy.

Proximity The target is near a node or zone. This is simple and durable, but it may only support room, aisle, sector, or boundary labels.

Signal strength Received signal level can suggest distance, but walls, orientation, body shielding, and multipath can move the estimate.

Time or angle Timing and angle measurements can improve geometry when clock, antenna, bandwidth, and calibration assumptions are visible.

Fusion Multiple observations can improve a record only when conflicts, stale data, and sensor quality are preserved rather than averaged away.

19.9.1 Targets and Association

Multiple targets add a new review problem: data association. When several measurements arrive, the tracker must decide which measurement belongs to which target. If two targets pass close together, a weak association rule can swap identities, merge two paths, or split one target into several tracks.

Scenario	Dominant review question
Single target	Does the record preserve energy-aware observation, localization, handoff, and recovery evidence?
Multiple targets	Does each measurement stay attached to the correct target identity or carry an uncertainty label?
Coverage gap	Does the page label the broken continuity instead of stitching two disconnected fragments into one certain track?

For multi-target cases, preserve identity evidence as carefully as position evidence. A useful record keeps the target id, candidate measurements, rejected associations, confidence boundary, and release label together.

19.10 Knowledge Check: Data Association

This section: Knowledge Check: Data Association

19.11 Handoff and Recovery Gates

This section: Handoff and Recovery Gates — At this point, handoff and recovery gates needs an inspectable chain rather than another rule of thumb.

At this point, handoff and recovery gates needs an inspectable chain rather than another rule of thumb. The figure at Figure 19.3 supplies it by linking Observation to source, age,.

WSN tracking fundamentals release gates showing observation evidence, prediction boundary, next owner, confirmation, recovery, gateway custody, energy state, and retest trigger. — Figure 19.3: WSN tracking fundamentals release gates for handoff, recovery, custody, and state labels.

Treat Observation, source, age,, and quality in the diagram at Figure 19.3 as three separate gates. The first identifies the field observation, the second establishes the starting condition, and the third marks the acceptance limit. Their combined message is specific: WSN tracking fundamentals release gates showing observation evidence, prediction boundary, next owner, confirmation, recovery, gateway custody, energy state, and retest trigger. Use that message, with its limits, in the ensuing handoff and recovery gates decision.

A handoff is not just a routing event. It is the transfer of tracking responsibility. The receiving node group needs the last observation, the prediction boundary, the confidence label, and the action to take if the target does not appear.

Before handoff Show the last confirmed observation, target state age, current owner, expected movement, and the evidence sent to the next owner.

During handoff Label whether the track is current, predicted, delayed, or ambiguous while the next sensor group waits for confirmation.

After handoff Confirm, correct, split, merge, or retire the state. Do not silently keep displaying a target that was never observed.

Recovery If the target is missing, show the search boundary, active sensors, timeout, and final lost or recovered label.

WSN tracking narrated tour

Narrated audio tour. The full transcript is below.

Play the tour to open the simulator. Pause, replay a beat, or read the full transcript below.

Some beats explain situations this teaching model cannot show; those beats keep the existing frame and provide captions only.

Full transcript — WSN tracking (8 beats)

B1 — Safety desk
You are at the safety desk while a lone worker inspects a large plant. Sensors detect the worker's badge near a pump, then another group reports a weak trace near the next room. Your decision matters: does the tracking history still describe the worker's current location?

Model caption: You staff the safety desk. A worker's badge observations must support a current location.

B2 — Observation evidence
A wireless sensor network connects observations over time. Detection means a sensor observed the target. Tracking means maintaining a labeled location history as the target moves. Keep the observation time with the location history. A prediction also needs an age, so the safety desk knows how old its supporting observation is.

Model caption: Detection is an observation. Tracking connects observations over time. Preserve observation time and prediction age.

B3 — Prediction and handoff
The worker keeps moving while some sensors sleep. Prediction estimates a future position and helps wake sensors near the expected next position. A handoff transfers tracking responsibility to the next sensor group. The receiving group needs the last observation and the uncertainty attached to it while waiting for confirmation.

Model caption: Prediction prepares likely next sensors. A real handoff carries the last observation and uncertainty to the next owner.

B4 — Detection lost
Use the teaching model's Recovery test: twelve sensors, a detection radius of fifteen metres, and a prediction lead of eight model steps. The target speed multiplier is one point six. Detection uses an ideal circular limit. After six advances, no sensor detects the target, and the lost count reaches one.

Model caption: Recovery test: 12 sensors; radius 15 m; speed 1.6x; prediction lead 8 steps. Step 6: 0 detecting, 1 lost frame.

B5 — Recovery radius
To find the recovery search radius, start with the count of consecutive advances without a detection, or zero, whichever is greater. For one missed detection, multiply one by two point two to get two point two metres. Add the fifteen metre detection radius to get seventeen point two metres. Add the model's fixed allowance of eight metres to get twenty five point two metres. The limit is forty two metres. Twenty five point two metres stays within the limit, so search within that distance of the predicted position.

Model caption: Search radius = min(42, 15 + max(0, 1) × 2.2 + 8) = 25.2 m around predicted position.

B6 — Detection resumes
On the next advance, a sensor detects the target again, and the lost count returns to zero. Recovery search can wake more sensors, adding energy cost. The model's energy figure counts awake sensor time only. Its detection and energy figures describe the teaching model, rather than field measurements about your worker.

Model caption: Step 7: 1 detecting, 0 lost frames. Detection resumes. Energy savings count awake sensor time only; model figures are not field evidence.

B7 — Forecast versus detection
A common mistake is to assume a handoff prediction confirms the worker's location, but prediction bridges a gap between observations; it does not refresh an old observation. During a handoff without confirmation, keep the uncertainty with the location history. The safety desk needs to distinguish a forecast from a fresh detection.

Model caption: Misconception: prediction proves a current location. An old observation stays old; preserve uncertainty until confirmation.

B8 — Fresh evidence
For your worker, a forecast is not a fresh detection.

Model caption: A forecast is not a fresh detection.

19.11.1 Prediction and Sampling Discipline

Detection and localization tell the reviewer where the target was at the last sample. In a duty-cycled WSN, the target keeps moving while nodes sleep, radios wait, and gateway delivery happens. Prediction bridges those gaps by waking likely next-zone nodes and by carrying a bounded future state until a new observation confirms, corrects, or retires it.

Prediction must still be labeled as prediction. A Kalman filter, particle filter, nearest-neighbor update, or simple dead-reckoning step can improve the next wake decision, but none of them turns a stale observation into a fresh one. The tracker should carry age, confidence boundary, missed-detection count, and retirement rule alongside the state.

The same fields protect energy-aware tracking. Selective activation saves energy only if sleeping zones, late wake commands, missing observations, and recovery-search costs remain visible in the evidence record.

19.12 Knowledge Check: Prediction and Sampling

This section: Knowledge Check: Prediction and Sampling

19.13 Energy State Is Evidence

This section: Energy State Is Evidence — Target tracking often reduces energy by activating only part of the network.

Target tracking often reduces energy by activating only part of the network. That is acceptable only when energy state is visible in the tracking record. A sleeping node cannot confirm a target. A recently awakened node may miss early evidence. A relay path may spend more energy than the sensing task.

Sleeping nodes State which zones were not observing and what that means for absence claims.

Wake timing Show whether nodes were awake before the target arrived or only after a prediction or query.

Communication cost Include reports, summaries, acknowledgments, retransmissions, and gateway handoff in energy discussion.

Recovery cost Lost-target search is part of the energy budget, not an exception outside the design.

19.14 Gateway Custody

This section: Gateway Custody — The target record may pass from sensor nodes to a cluster head, mobile collector, gateway, cloud service, dashboard, or operator.

The target record may pass from sensor nodes to a cluster head, mobile collector, gateway, cloud service, dashboard, or operator. Each custody step can change what the record means.

Local custody Nodes and clusters preserve raw or summarized observations while the target is near them.

Network custody Routing, aggregation, and store-forward behavior can delay, combine, or drop updates.

Application custody Dashboards and alerts must label whether a state is current, predicted, delayed, partial, or historical.

Audit custody A later reviewer should be able to reconstruct why a target state was displayed or retired.

19.15 Worked Review: Boundary Motion

This section: Worked Review: Boundary Motion — Scenario: a perimeter WSN detects motion near a boundary and shows a target marker moving toward a restricted zone.

Scenario: a perimeter WSN detects motion near a boundary and shows a target marker moving toward a restricted zone.

Claim to review The network is tracking a moving target, not merely detecting one motion event.

Evidence required At least one confirmed observation, a current state age, a next-zone prediction, and a handoff or recovery plan.

Risk label If only one sensor fired, label the marker possible or ambiguous until another observation confirms continuity.

Approval decision Approve only if the page shows observation, estimate, handoff, gateway receipt, and release label.

19.16 Worked Review: Asset Lookup

This section: Worked Review: Asset Lookup — Scenario: an operator asks where a tagged tool was last seen in a facility.

Scenario: an operator asks where a tagged tool was last seen in a facility.

Claim to review The system can report a last-known zone and the age of that evidence.

Evidence required Query scope, responding nodes, missing nodes, last observation time, and whether the tag was moving.

Risk label A no-response result is not proof of absence. It is a bounded query result with missing coverage if nodes did not answer.

Approval decision Approve a historical or delayed label if the page clearly separates last-known evidence from current location.

19.17 Worked Review: Mobile Collector

This section: Worked Review: Mobile Collector — Scenario: a mobile collector receives target summaries from a sparse sensor field while moving past several zones.

Scenario: a mobile collector receives target summaries from a sparse sensor field while moving past several zones.

Claim to review The collector can carry custody of target records without making stale summaries look current.

Evidence required Collection time, sensor time, relay delay, path segment, dropped summaries, and whether follow-up confirmation occurred.

Risk label If gateway contact happens after the target moved, the released state should be historical or delayed.

Approval decision Approve only when the record preserves both observation time and delivery time.

19.18 Common Tracking Mistakes

This section: Common Tracking Mistakes

Turning detection into certainty One motion event, packet, or signal sample becomes a target track without geometry, confirmation, or age.

Hiding handoff gaps The page shows a smooth path even though no sensor owned the track during a transition.

Using prediction as proof A predicted marker is displayed as current even after its supporting observation is stale.

Ignoring sleeping zones The network claims absence in areas that were not awake, not queried, or not reachable.

Dropping custody history Gateway delays, aggregation, and store-forward behavior are hidden, making old evidence look fresh.

Overstating accuracy The chapter gives fixed accuracy or battery-life claims without deployment measurement, geometry, or failure labels.

19.19 Readiness Checklist

This section: Readiness Checklist

Evidence Observation source, timestamp, measurement quality, localization method, and uncertainty are visible.

Continuity Handoff, owner, prediction, confirmation, loss, and recovery behavior are visible.

Energy Sleeping, waking, sampling, relaying, gateway handoff, and recovery are included in the tracking claim.

Release The final target state is labeled as current, predicted, delayed, ambiguous, lost, recovered, partial, or historical.

19.20 Knowledge Check: Continuity

This section: Knowledge Check: Continuity

19.21 Knowledge Check: Measurement Evidence

This section: Knowledge Check: Measurement Evidence

19.22 Matching: Tracking Vocabulary

This section: Matching: Tracking Vocabulary

19.23 Ordering: Tracking Record Review

This section: Ordering: Tracking Record Review

19.24 Summary

This section: Summary — WSN tracking fundamentals are about preserving target meaning over time.

WSN tracking fundamentals are about preserving target meaning over time. A strong chapter separates detection from tracking, observation from prediction, and current state from stale or historical state. It makes handoff, uncertainty, energy state, custody, and recovery visible before asking learners to trust a target marker.

The practical review question is simple: can another person reconstruct why the page displayed this target state? If the answer is yes, the tracking record is ready for formulation, algorithm, prediction, implementation, and application-specific review.

19.25 Key Takeaway

This section: Key Takeaway — WSN Tracking Fundamentals Review should connect tracking models, prediction, localization, handoff, sampling rate, energy cost, uncertainty, and deployment evidence before accepting a design.

WSN Tracking Fundamentals Review should connect tracking models, prediction, localization, handoff, sampling rate, energy cost, uncertainty, and deployment evidence before accepting a design.

19.26 Concept Relationships

This section: Concept Relationships

From coverage WSN Coverage Fundamentals explains whether sensors can observe the target area in the first place.

From routing WSN Routing Fundamentals explains how target evidence reaches cluster heads, gateways, and applications.

To formulations WSN Tracking Formulations Review chooses push, poll, guided, or hybrid behavior after the fundamentals are clear.

19.27 What’s Next

This section: What’s Next — Use WSN Lab Exercises to connect tracking claims to repeatable evidence records.

Previous: WSN Lab Exercises

Next: WSN Tracking Formulations Review

Use WSN Lab Exercises to connect tracking claims to repeatable evidence records. Continue to WSN Tracking Formulations Review for push, poll, guided, and hybrid tracking behavior.

Choose the tracking behavior Continue to WSN Tracking Formulations Review for push, poll, guided, and hybrid formulation choices.

Review algorithm components Use WSN Tracking Algorithm Components Review to inspect detection, cooperation, localization, prediction, and recovery.

Review selective activation Use WSN Tracking Energy Prediction Review when the tracking claim depends on prediction-driven wake zones.

Find help

Search reviewed answers

3 reviewed answers.

Where should I start in a module?

Open the module page from the sidebar to see its chapters in order. Use the module map to return to a topic you need.

Modules and paths

Reviewed 2026-10-16

What can I try when a project does not work?

Check the symptom, then work through the diagnostic steps. Keep the error message and the smallest example that reproduces it.

When you are stuck

Reviewed 2026-10-16

How do interactive examples work?

Use the interactive guide to check how examples and simulations work before changing your project.

Interactive examples guide

Reviewed 2026-10-16