This project consisted of applying Dynamic Power Management (DPM) on a series of worloaks by simulating and modifying a Power State Machine (PSM) written in C language. The manager implements two strategies:
-
Timeout Policy, that keeps the system in a high power consuming Run state even after activity has ceased, transitioning to a low power state only after a certain threshold is reached, i.e., the timeout value.
-
History Policy, that is a predictive policy, i.e., it estimates the time to transition based on the previous workload patterns. The computed value Tpred is compared to the time to Idle and Sleep thresholds. After a decision is taken the system is able to reach any one of the low-power states independently.
Evaluation of several image transformations used in Organic Light-Emitting Diode (OLED) displays as a mean of reducing power consumption using MATLAB functions. Compensation techniques are also used in conjunction with Dynamic Voltage Scaling (DVS) for the same goal.
Simulation of an IoT device in MATLAB/Simulink. The model consisted of a System-on-Chip (SoC) for measuring urban polution, integrating an the following sensors: air quality, methane level, temperature, and noise. A subsequent analysis of its behavior in terms of power perspective and possible optimizations to increase the system lifetime are provided.
Further explanations can be found in the file "report.pdf" located inside each respective folder.