397  WSN Tracking: Interactive Demo and Review

397.1 Learning Objectives

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

  • Experiment with Tracking: Use interactive simulations to understand sensor activation and target tracking
  • Test Understanding: Answer knowledge check questions covering tracking formulations, algorithms, and trade-offs
  • Visualize Concepts: See how detection zones, estimated positions, and predictions work in real-time
  • Apply Knowledge: Connect theoretical concepts to practical tracking scenarios
TipMVU: Minimum Viable Understanding

Core concept: Interactive experimentation reinforces theoretical understanding of WSN tracking concepts. Why it matters: Seeing tracking algorithms in action reveals nuances that static diagrams cannot convey. Key takeaway: Move the target through the sensor field to observe detection, estimation, and energy savings in real-time.

397.2 Prerequisites

Before diving into this chapter, you should be familiar with:

397.3 Interactive Target Tracking Demo

Experience WSN target tracking in action! This simulation demonstrates how sensors detect and track a moving target across a 2D deployment area.

How to use: - Drag the target (orange circle) to manually move it, or enable Auto-Move - Watch sensors light up green when they detect the target (within detection radius) - Enable Prediction Mode to see where the target is heading (blue dashed line) - Observe estimated position (red cross) vs actual position (orange circle) - Monitor tracking accuracy and active sensors metrics in real-time

Key observations:

  1. Detection zones (dashed circles) show each sensor’s coverage area
  2. Active sensors (green) are those currently detecting the target
  3. Estimated position (red cross) is calculated using weighted trilateration
  4. Tracking accuracy decreases when fewer sensors detect the target
  5. Energy savings shows what percentage of sensors are sleeping (not needed for tracking)
  6. Prediction (blue arrow) shows anticipated target trajectory based on velocity

Try these experiments:

  • Move target to an area with only 1 sensor - notice the high estimation error
  • Move target to center where 3+ sensors overlap - observe improved accuracy
  • Watch how prediction adapts when you change target direction
  • Compare energy savings at edges (fewer sensors active) vs center (more sensors needed)

397.4 Knowledge Check

Test your understanding of WSN tracking fundamentals with these questions.

Question: In push-based target tracking, sensors proactively report detections to a base station. In poll-based tracking, the base station queries specific sensors. What is the primary trade-off between these approaches?

Explanation: B. Push minimizes detection-to-sink latency but increases transmissions; polling reduces transmissions but adds query latency and can miss short events.

Correct: B) Push-based enables real-time tracking with higher energy cost; poll-based is energy-efficient but adds query latency

Push-based tracking: - Proactive reporting: Sensors detecting target immediately transmit position estimates - Real-time: Minimal latency from detection to base station awareness (milliseconds to seconds) - Higher energy cost: More frequent transmissions from detecting sensors

Poll-based tracking: - On-demand queries: Base station polls specific sensors only when a position update is needed - More energy-efficient: Sensors can sleep until polled, transmitting only when queried - Added latency: Query propagation + sensor measurement + response transmission = seconds to minutes

Message-count comparison (illustrative): - Push-based (1 report/min): ~43,200 uplinks per node per 30 days - Poll-based (1 query/hour): ~720 uplinks per node per 30 days - That’s ~60x fewer uplinks (often the dominant energy difference), though real battery life depends on payload size, radio current, and downlink cost.

Question: How many sensors with distance measurements are required (minimum) to determine a unique 2D position using trilateration?

Explanation: C. Two ranges leave two candidate intersections; a third range usually selects the valid point (noise-free geometry).

Correct: C) 3 sensors - three distance circles have a unique intersection point

Trilateration mathematics: - 1 sensor: Distance d1 from sensor at (x1, y1) defines a circle - infinite points possible - 2 sensors: Two circles intersect at exactly 2 points (ambiguity remains) - 3 sensors: Three circles have a unique intersection point (in ideal case)

Trilateration equations:

(x - x1)² + (y - y1)² = d1²
(x - x2)² + (y - y2)² = d2²
(x - x3)² + (y - y3)² = d3²

In practice (4+ sensors preferred): - Real measurements have noise and errors - 4+ sensors enable least-squares optimization - Overdetermined system provides error correction and outlier rejection - Geometric Dilution of Precision (GDOP) improves with strategic sensor placement

Question: A Kalman filter is used to estimate target position and velocity from noisy measurements. What key advantage does it provide over simple averaging?

Explanation: B. The filter fuses a prediction (motion model) with new measurements, improving stability and enabling look-ahead activation.

Correct: B) Kalman filter uses a motion model to predict future positions, enabling smoother tracking and proactive sensor activation

Kalman filter key components:

  1. State vector: Contains position (x, y) AND velocity (vx, vy)
  2. Motion model: Predicts next state based on physics (x’ = x + vx*dt)
  3. Prediction step: Estimates where target will be before next measurement arrives
  4. Update step: Corrects prediction when actual measurement received

Advantages over simple averaging: - Estimates velocity (not just position) - enables trajectory prediction - Optimally weights prediction vs measurement based on uncertainty - Handles irregular measurement timing gracefully - Provides uncertainty estimate (covariance) for each position

Energy efficiency impact: - Predicted position enables proactive sensor wake-up along expected path - Sensors NOT on predicted path can remain sleeping (85-90% energy savings) - If target moves unpredictably, expanding ring search recovers tracking

Question: When multiple sensors detect the same target, what is the purpose of using RSSI-weighted averaging during fusion?

Explanation: B. RSSI is a noisy proxy for link quality/distance; weighting can reduce the influence of far or obstructed sensors in the fused estimate.

Correct: B) Stronger RSSI indicates closer proximity, so those measurements should have higher weight in position calculation

RSSI-weighted fusion rationale: - Stronger RSSI (e.g., -50 dBm) often indicates a closer or less obstructed link (but RSSI can be noisy due to multipath) - Weaker RSSI (e.g., -80 dBm) often indicates a farther or more obstructed link, increasing variance - Weighting: w_i = 1 / (distance_i + epsilon) gives closer sensors higher influence

Path loss model (RSSI to distance):

d = d0 * 10^((RSSI0 - RSSI) / (10 * n))

Where RSSI0 is the RSSI at reference distance d0 (often 1m), and n is the path loss exponent (approximately 2 in free space, 3-4 indoors)

Example fusion: - Sensor A: RSSI = -50 dBm (estimated distance: 2m) - weight = 0.5 - Sensor B: RSSI = -70 dBm (estimated distance: 10m) - weight = 0.1 - Sensor A’s measurement contributes 5x more to final position estimate

Cluster head aggregation: - Nearby sensors report to cluster head - Cluster head performs weighted fusion - Single fused position forwarded to sink (reduces multi-hop transmission overhead)

Question: When a tracked target is lost (prediction failed and no sensors detect it), what is the expanding ring search strategy?

Explanation: B. It prioritizes likely locations first, limiting how many sensors you wake while still guaranteeing coverage as the radius expands.

Correct: B) Start searching in a small radius around last known position, then progressively expand the search area based on maximum possible target speed

Expanding ring search algorithm:

  1. Last known position: Target was at (x0, y0) at time t0, then tracking lost
  2. Maximum displacement: Target cannot travel faster than v_max
  3. Search radius: After time dt, target is within radius r = v_max * dt
  4. Progressive activation: Activate sensors in expanding rings

Algorithm steps:

for radius in range(10, max_radius, step=10):
    # Activate sensors within current ring
    ring_sensors = sensors within [radius-10, radius] from last_position
    for sensor in ring_sensors:
        if sensor.detect_target():
            return sensor.position  # Target recovered!

Energy efficiency: - Ring search activates ~10-15 sensors per ring (annular area) - If target nearby (common case): found in 1-2 rings, minimal energy - Full network activation (all 100 sensors) only if target at boundary

Recovery time calculation (illustrative): - Target moving at 2 m/s, lost for 30 seconds - Maximum displacement: 60m - If you expand by 10m per step, recovery could take ~6 steps in the worst case

397.6 Summary

This chapter provided hands-on experience with WSN tracking concepts:

  • Interactive Demo: Experimented with real-time target tracking, observing how sensor detection zones, position estimation, and energy savings change as the target moves.

  • Trilateration Visualization: Saw how position accuracy improves with more detecting sensors (1 sensor = high uncertainty, 3+ sensors = trilateration precision).

  • Prediction Visualization: Watched the prediction arrow adapt to target velocity changes, understanding how this enables proactive sensor activation.

  • Knowledge Checks: Tested understanding of tracking formulations (push vs poll), trilateration requirements, Kalman filtering benefits, RSSI-weighted fusion, and expanding ring recovery.

397.7 What’s Next

Continue exploring WSN tracking with these related chapters: