A Novel Design Framework for Embedded Applications
Abstract
Real-time microcontrollers are typically constrained to different performance parameters including code size, execution time, and power consumption. These system constraints cooperate with each other in a complicated way, making it challenging to develop an optimized design methodology for real-time embedded systems. In this work, we propose a design framework by reviewing the tradeoff between code size and execution time. We applied Period Calibration Method (PCM), which converts temporal system constraints into task parameters. PCM derives temporal parameters and code size parameter of each task, and thus determines the system end-to-end timing requirements with reduced code size. The proposed methodology makes a set of design parameters to optimize the real-time embedded system output. Our design framework uses Earliest Deadline First (EDF) scheduling technique to improve system utilization by reducing dead cycles. The level of optimization is evaluated using ARM7 development suite. Dynamic Voltage Frequency Scaling (DVFS) method is applied to optimize power consumption by adjusting the frequency. The proposed design framework results are potential. The proposed optimization techniques help optimize the code size up to 21.22%, execution time up to 23.61%, and frequency up to 31.75%. This work can be extended to conduct power optimization of various microcontroller systems.