position and velocity of the tag. The position of the tag at a specific time is the position last time we measured to which is added how far the tag has travelled since then with a velocity driven by a white noise process. The observation equation (nonlinear) relates the position of the tag to the received signal strength at the APs measured with some noise. Due to the way the Bluetooth protocol is specified RSSI from one tag may not necessarily be sampled at the same time at other APs. Therefore a preprocessing of data is used to align the measurements in time. This is done using a Kalman filter on a local level model [3] to smooth and align the data. Using the extended Kalman filter on the preprocessed data gives an estimate of the route of the tag. The positions on the route together with the information from the lying/standing tag may be used to calculate different measures of activity to be used in subsequent modelling. This talk will present the positioning algorithm and the results from the algorithm to be used in calculating some activity measures.
[1]
Andrew Harvey,et al.
Forecasting, Structural Time Series Models and the Kalman Filter
,
1990
.
[2]
Richard A. Brown,et al.
Introduction to random signals and applied kalman filtering (3rd ed
,
2012
.
[3]
S. Kerbrat,et al.
A proposition for an updated behavioural characterisation of the oestrus period in dairy cows
,
2004
.
[4]
B. Hart,et al.
The behavior of sick animals.
,
1991,
The Veterinary clinics of North America. Small animal practice.