Power-Aware Data Structure Using Memory Allocation Techniques in EPS-Based Motion Recognition Systems

摘要

In this paper, we propose power-aware data structure using loop transformation in preprocessing of motion recognition systems based on EPS (Electromagnetic Potential Sensor). First, we estimate power consumption of smart devices or tablets in gesture recognition preprocessing. We analyze input signals from sensors, filter, store, and measure the power efficiency of each of different data structures using loop transformation. We search data structure for optimal transformation. Proposed method based on EPS for gesture recognition in smart devices demonstrates to reduce power consumption significantly, about 20%.