Energy Saving Using Scenario Based Sensor Selection on Medical Shoes

Cost and energy have become the bottleneck of many medical embedded systems. It is mainly due to the fact that wireless medical devices used in wireless health applications normally employ a large number of sensors, which are both expensive as well as consuming a considerable amount of energy. In this paper, we have developed a cost and energy saving scheme by reducing the number of required sensors in the medical devices. We have used a popular medical shoe with 99 pressure sensors for demonstration. With the goal of reducing the number of required sensors without influencing the diagnose accuracy, we have proposed algorithms to select only a small subset of the sensors while still maintaining an almost same diagnostic performance compared to using all the 99 sensors. Our results indicate that on average, our sensor selection can save as much as 88% of the total sensors. We also analyze the effect of using the medical shoes under different scenarios on the results of sensor selection. For example, we have analyzed the scenarios of walking, jumping, running and slow walking. Based on our sensor selection algorithm, it turns out that there exists different subsets of sensors to best recover the diagnostic performances for different scenarios.