Dementia is highly prevalent among the older population. Most patients with dementia are admitted to an aged care facility due to wandering behaviour which tends to result in dangerous scenarios such as straying away from the facility and being seriously injured. Due to the decreasing availability of carers in aged care, there is a need to prioritise monitoring of patients that have a severe case of wondering. The challenge is to allow carers to monitor the status of such patients in terms of position localisation and motion behavioural status, in real-time. The long term behavioural analysis of such patients would allow carers to better manage such patients. Current indoor localisation technologies cannot provide the accuracy of location and motion to enable unobtrusive behavioural analysis. Our aim was to develop an indoor localisation and activity monitoring system for aged care workers to aid the prioritisation of surveillance to the patients with dementia. Our system used Radio Frequency tracking combined with motion and heading sensors to track a person. The motion and heading sensor information were incorporated into a human activity classification model to determine the characteristics of a patient's walking activity. We conducted a month-long trial of our localisation network and activity monitoring system in an aged care facility.