International Biometric Society RSSI based indoor tracking in sensor networks using extended Kalman filters

The first symptom of a disease of a cow is very often a change in behavioural patterns related to reduced activity, decreased feed intake, and increased time spent lying [2]. Increased activ-ity such as increased walking together with reduced time spent on eating and lying is typical indicators for cows in heat [4]. To monitor the behavioural patterns of cows in order to detect diseases and oestrous of cows a wireless sensor network with access points (APs) based on the Bluetooth protocol has been established in the barn at the Danish Cattle Research Cen-tre. This network is able to communicate with several devices (tags) mounted on the cows and data from the tags is stored on a central data server. A tag mounted on the cow’s leg measures whether the cow is lying or standing. The received signal strength indicator (RSSI) between tags and APs are recorded over time using a tag mounted on the back of the neck of the cow. The positions and walking behaviour of a cow may be estimated using the measures of RSSI. Supplementary data from feeding stations and milking robots form together with the sensor network data the basis on which biometric models of behavioural patterns is currently being developed. Here we discuss the method currently being implemented to estimate the positions of cows. The RSSI from a tag recorded with timestamps at each AP is used to es