Abstract
In real-time embedded systems, scheduling policy is considered one of the main factors that affect their performance. It helps to choose which task should be selected first from ready queue to run. Round Robin (RR) scheduling algorithm is widely used and its performance highly depends on a Quantum size Qt, which is a predefined amount of time assigned by CPU to every task to be executed. However, the performance degrades with respect to an average waiting time (A WT), an average TurnAround time (ATT) and a number of Context Switches (NCS). This paper presents a new improved dynamic Round Robin scheduling algorithm to reduce the average waiting time, turnaround time and the number of context switches in order to improve the system overall performance. It also presents a comparative analysis between several existing Round Robin algorithms based on the average time for waiting and turn-around and number of context switches.