Research On The Embedded Low Power Consumption Technology Customized For Wearable Applications

With the rapid development of wearable technologies,more and more wearable devices are entering people's daily life.Since wearable devices use internal batteries to provide power,how to ensure wearable device's performance while reducing the power consumption of the wearable device and improve the life of the wearable device has became the key issues to be studied.This article first analyzes the hardware and software platform of the actual wearable system,analyzes the running characteristics of the wearable application in detail,analyzes the main power consumption source of the wearable hardware platform through the experiments,and proposes a method of software and hardware cooperation to begin the low-power technology research.The detail of the new features of the EAS scheduler is analyzed.Combined with the actual wearable application scenario,an improved scheduling scheme for wearable applications is proposed.Aiming at the shortcomings of using the mean value to predict the task load in the WALT algorithm,the "exponential smoothing-based WALT algorithm" is proposed to improve the accuracy of the task load forecasting;The CPU selection algorithm is improved,which selects CPU cores with higher energy efficiency for delay-sensitive tasks and non-delay-sensitive tasks respectively.The frequent migration of associated tasks on the CPU cores is avoided,which reduces the waste of power.In response to the problem of global load balancing,this article proposes the "Favored Partial Load Balancing Strategy",which improves the efficiency of the load balancing.The shortages of CPUFreq frequency modulation module and CPUIdle sleep module are analyzed.The power management optimization strategy for wearable applications is proposed.The adjustment strategy of CPU frequency is optimized,and the CPU frequency is avoided to switch back and forth.At the same time,if the system is idle,let the system get into a deep sleep state,thereby reducing the overall power consumption of the system.Finally,a contrast experiment was conducted on the scheduling strategy and power management before and after the improvement.Experimental result shows that,in terms of basic performance,the improved scheduler has about 5% improvement.In the multi-threaded benchmark test programs,the improved scheduler ensures that the performance is basically the same,while the power consumption has about 5% improvement.And in actual wearable application scenario testing,the improved scheduler has about 8% improvement in power consumption.