← Back to the chapter · Previous detail · Next detail →

Technical detail · IoT Mathematics: Fusion and Sampling Decisions

Measurement weighting and uncertainty

Step 3: Measurement Update (GPS Reading)

GPS reports position (103m, 201m) with sigma = 10m uncertainty.

From Section D.2 (Gaussian Distribution), GPS measurement has:

From Section D.3 (Sensor Fusion), combine prediction with measurement:

pfused=σGPS2⋅ppred+σpred2⋅pGPSσpred2+σGPS2p_{fused} = \frac{\sigma_{GPS}^2 \cdot p_{pred} + \sigma_{pred}^2 \cdot p_{GPS}}{\sigma_{pred}^2 + \sigma_{GPS}^2}

Assume prediction uncertainty σpred=5m\sigma_{pred} = 5m (accelerometer integration), GPS uncertainty σGPS=10m\sigma_{GPS} = 10m:

X-coordinate: px=100⋅101.5+25⋅10325+100=10150+2575125=12725125=101.8mp_x = \frac{100 \cdot 101.5 + 25 \cdot 103}{25 + 100} = \frac{10150 + 2575}{125} = \frac{12725}{125} = 101.8m

Y-coordinate: py=100⋅200+25⋅201125=20000+5025125=200.2mp_y = \frac{100 \cdot 200 + 25 \cdot 201}{125} = \frac{20000 + 5025}{125} = 200.2m

Fused estimate: (101.8m, 200.2m)

Notice the fused position is closer to the prediction (101.5, 200) than the GPS (103, 201) because prediction is 2x more accurate (σ=5\sigma=5 vs σ=10\sigma=10).

Step 4: Uncertainty Reduction

From Section D.3, combined variance:

σfused2=σpred2⋅σGPS2σpred2+σGPS2=25⋅10025+100=2500125=20\sigma_{fused}^2 = \frac{\sigma_{pred}^2 \cdot \sigma_{GPS}^2}{\sigma_{pred}^2 + \sigma_{GPS}^2} = \frac{25 \cdot 100}{25 + 100} = \frac{2500}{125} = 20

σfused=20=4.47m\sigma_{fused} = \sqrt{20} = 4.47m

Interpretation: Fusing 5m-accurate prediction with 10m-accurate GPS yields 4.47m accuracy - better than either sensor alone!

← Back to the chapter · Previous detail · Next detail →