A single performance measure for perimeter intruder detection systems

Evaluations of perimeter intruder detection systems (PIDS) often contain results for probability of detection (P_d), false alarm rate (FAR), and costs. In a comparative evaluation it is difficult to choose the best of several systems when presented with several performance measures for each solution (e.g. P_d, FAR and cost). P_d may be best for one system, FAR for another, and cost for a third. It is not immediately clear how these performance measures should be traded-off. Those purchasing PIDS would like to use performance-based procurement to obtain a system with the best P_d vs FAR tradeoff, and allow tender selection other than on the basis of least capital cost. Existing performance specifications contain selection criteria that are somewhat arbitrary. For example, a selection criterion might be that the detection rate must be demonstrated as at least 90%. This implies that 89% is a complete failure, and that no reward is given if a detection rate of 95% is demonstrated. This is not what the specifier actually wishes to indicate. The paper describes ways to portray the comparative performance of PIDS and proposes a way to combine P_d, FAR, reliability figures and costs to produce a single, net value for each system´s performance. It is suggested that this value could be used to: (a) rank PIDS evaluation results, (b) help with tender selection, and (c) suggest any final performance-linked payment