| Embedded system design requires frequent analysis and verification of new features to ensure correctness and effectiveness; this becomes extremely difficult while hardware systems are turning increasingly complex with the development of embedded technology. Traditional methods such like constructing a physical system for each of the new features are no longer feasible due to cost and timeline limits; and simple analysis and modeling methods are not reliable because of low precision and narrow scope of application.By simulating the structure and behavior of the system, architecture simulator can achieve the purpose of system analysis and verification. In order to use simulation technology to analyze and verify the system, a base system and its measured simulator are needed. Starting from this measured simulator, merge the new designs into it one by one, and a new system is achieved. Therefore, the measurement between the base system and the initial simulator is critical. In addition, optimization techniques are important to simulation technologies, since the simulator needs to run fast and with high accuracy in order to get effective data within an acceptable time period.Current research focus on the design and development of simulator, and simulator calibration and measurement mainly depends on developer's experience; the lack of theoretic support can reduce the performance and accuracy of simulator.This paper proposed a set of measurement methods for calibration of real systems and simulators. Also, several optimization methods are introduced. Experiment shows that these methods can achieve high accuracy and improve the simulator's performance. |