Signal Processing of our new Enhanced HasMoved Value

Since day one our ZIGPOS Badge comes with an acceleration sensor (the ST LSM9 series) on board to recognize movement of the Badge for high precision positioning. When no movement is detected, the Badge is in sleep mode to save energy. This is a key functionality to energy efficiency.

Starting with our newest firmware, we are introducing the Enhanced HasMoved Value, which is an indicator to estimate how massive the acceleration of the Badge was.

To keep the channels free and to be able to scale the user scenarios to 1000+ Badges, we are not interested in transmitting the raw sensor data over the air. Instead we are preprocessing on the Badge. The following signal processing is done on the Badge.

Process Raw Accelerations to get the Enhanced HasMoved Value

The ST 9DoF IMU provides acceleration values for each of the three cartesian acceleration vectors. We sample these with \(f=50Hz\).

In the following figure, one can see these 3 acceleration sensor values with some characteristic noise and some movement of the Badge between second 3…9s.

The Signal Processing is done embedded and energy efficient in three steps:

  1. calculate the linear acceleration of all 3 axis without gravity (\(|a|=\sqrt{a_x^2+a_y^2+a_z^2}-g\))
  2. sample & hold 5 values of \(|a|\) and calculate the mean of these 5 \(\hat{|a|}_5=\frac{|a|_{t-4}+|a|_{t-3}+|a|_{t-2}+|a|_{t-1}+|a|_{t}}{5}\)
  3. calculate the variance of this \(\hat{|a|}_5\) vector over 1 second of time

These 3 steps can be seen in the following three plots (from top to bottom):

This is resulting in the Enhanced HasMoved Value (ehv), which is transmitted to the ZIGPOS Gateway and can be used to adapt tracking filters like the Kalman or Particle filter.

Get in touch with our sales team to learn more about the Enhanced HasMoved Value and how you can integrate it in your application.

Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert.

You are here

Latest Tweets