Cattle monitoring system using wireless sensor network in order to prevent cattle rustling

Stock theft is a major problem in the agricultural sector in South Africa and threatens both commercial and the emerging farming sectors in most of the country. Although there have been several techniques to identify cattle and combat stock theft, the scourge has not been eradicated in the farming sector. This paper investigates how we can model cow behaviour using global positioning wireless nodes to get the expected position of a cow. The objective of this research is to model the typical behaviour of a cow to determine anomalies in behaviour that could indicate the presence of the thieves. A wireless sensor node was designed to sense the position and speed of a cow. The position and the speed of the cow are collected for analysis. A random walk model is applied to the cow's position in order to determine the probability of the boundary condition where we assume there is an increased probability of a cow on the boundary position being stolen. The Continuous Time Markov Processes (CTMP) is applied to the movement pattern of an individual cow in order to find the probability that the cow will be at the boundary position. The value of 2.5 km/h has been found as our treshold to detect any agitation of the animal. The cow has less probability to be at the boundary position. The predictive model allows us to prevent stock theft in farms especially in South Africa and Africa in general.