An Algorithm for Improving Resolution of Microwave Resonant Sensors for Blood Glucose Monitoring

This paper discusses a method to improve the resolution of microwave biomedical sensors. In particular, the specific focus is on microwave sensors dealing with the problem of continuous blood glucose monitoring, in which the shift of the resonance frequency is tied to the variation of the glucose concentration. Usually, since the quality factor of such sensors is usually relatively low, it follows that the resonance peak is not very pointed. Furthermore, the sensor response is gathered over a finite and countable set of frequencies, and hence the actual resonance peak may not be properly caught. To overcome these drawbacks, a procedure based on a super-resolution method is presented, in order to accurately detect the resonance frequency. The algorithm is first evaluated on a synthetic function simulating the response of a real resonant sensor; subsequently, measured data collected by a patch resonant antenna which is immersed into a water glucose solution with different level of glucose concentrations are also considered.