Research On Status Monitoring For Embedded Software

With the continuous development of computer technology,embedded system has been applied widely in many fields,such as industrial,medical,communications and aerospace,etc.The increasing complexity of operating environment of embedded system makes the software system bigger and bigger. This makes the hidden trouble of the embedded system growing,and these hidden dangers may suddenly and violently expose during the runtime. The accident may cause unnecessary losses. This thesis research on the methods of monitoring the performance of VxWorks operating system.Make sure that system software aging caused by the performance and behavior disorders can be found in time.Firstly, this thesis study on the methods of monitoring embedded system performance.The real-time memory of running system, the real-time status running state of the task, the real-time CPU occupancy rate and the real-time stack usage of the task are monitored at first.At the same time, the embedded software’s aging trend is analyzed and the data about software aging is collected. Two methods of determine the software aging trend are described.Secondly, Hidden Markov Model and genetic algorithm are combined to monitoring the behavior of the embedded software. Due to the Baum-Welch is very sensitive to training parameters of HMM, and the selection of initial parameters are completely random, Hidden Markov Models will be trapped in local optimal problems.In this thesis,genetic algorithm is used to optimize the training process to form better Hidden Markov Model.The Hidden Markov Model can be used to monitoring the behavior of software.Thirdly, this thesis study on the methods of processing abnormal state of the embedded software. After analyzing the abnormal state of system, according to different anomalies, we use different strategies to deal with the anomalies. In order to improve the reliability of software,this thesis presents a decision model of software.At last,the processing result is evaluated.Finally, we conduct experiments to testify the proposed methods. Simulating the memory leaks, task anomalies and stack cross-border, which are often encountered in the actual operating system.Then, the real operation of the typical tasks are simulated, the task model is established. Collect the task running time, and according to the threshold calculation formula to calculate the time threshold. Finally, we use the AUV simulation software to verify the validity of the algorithm.