Biometric Sensor Management: Tradeoffs in Time, Accuracy and Energy

In this paper a novel sensor management algorithm is presented for a biometric sensor network. A distributed detection framework is managed for different energy, accuracy and time requirements. The design variables include sensor thresholds, fusion rule, sensor selection and sensor mode selection. Different sensors are associated with different transaction times. Hence, varying sensor modes can affect the accuracy and energy consumption. Once the sensors and their modes are selected, the accuracy achieved by this subset of sensors is maximized by managing the thresholds and the fusion rule. Risk, time and energy are the three objectives that the system attempts to minimize. The three objectives are tied into a single objective function by weighting them. A hybrid particle swarm optimization algorithm is design the system. The algorithm is a hybrid of continuous, discrete and binary particle swarm. The continuous particle swarm is used to manage the thresholds. The binary particle swarm is used to manage the fusion rule. The discrete particle swarm is used to select the sensors and the sensor mode. The system is adapted for different threat levels that depend on the a priori of imposter in the network. Results show the effectiveness of the proposed method in adapting the system to different requirements under different threat situations.