Age of Information Optimization Using a Hybrid Preemptive/Non-Preemptive Discipline

Currently, the pervasive adoption of internet of things technology sheds light on the significance of real-time status update systems. The age of information (AoI) metric has been devised to reflect the strict information timeliness of such systems. Bufferless packet management scheme has demonstrated higher effectiveness in the AoI minimization. However, the existing preemptive (PR) and non-preemptive (NP) service disciplines operate independently of the system state, thereby reducing the system’s adaptability to varying system parameter settings, such as service time distribution and traffic loading condition. Therefore, we propose an elapsed-time-based hybrid PR/NP discipline to govern the service preemptions in a single-source M/Er/1/1 queueing scheme. The preemption is declined when the elapsed service time surpasses a predetermined threshold (controlling parameter). The average AoI is analyzed using the stochastic hybrid system approach. The numerical study demonstrated that the proposed discipline, owing to its threshold parameter, fulfills the optimality of the AoI performance, especially under highly dispersed Erlang service time distributions and higher traffic loading conditions.