A reconfigurable hardware platform for cognitive sensor networks towards behavioral biometrics

This paper presents a design of reconfigurable hardware platform for wireless sensor networks to achieve high flexibility and robustness in system operation and resource utilization. The goal of this research is to develop a new type of sensor networks with cognitive features (e.g., context awareness and situation understanding) such that human behavioral biometrics can be measured through distributed low-cost sensors. The proposed reconfigurable hardware platform can adjust its sensing and computing resources with respect to changing contexts and situations. It includes three components: (1) reconfigurable analog circuit for adaptive sensor conditioning, (2) reconfigurable ADC for maintaining high signal-to-noise ratio, and (3) FPGA based reconfigurable computing for situation understanding. In behavioral biometrics applications, such a reconfigurable platform can automatically change circuit parameters, ADC resolutions, filter coefficients, and algorithms according to application contexts and situations derived from sensory signal features. As a result, the sensing system can be adaptive to various behavioral biometrics applications with high fidelity and accuracy. Experimental results have demonstrated the advantages of the proposed hardware platform.