Information Freshness in Sleep-Wake Server Systems

Information freshness has garnered wide attention due to its importance in various applications. In this work, we consider an information updating system in which the server falls asleep after a random idling period if no arrival occurs. The system is modeled as an M/G/1 queue with server sleeps, and we further derive closed-form expressions for Peak Age of Information (PAoI) of the system under three different sleeping strategies, namely, N-policy, single sleep, and multiple sleep. We then show both analytically and numerically that reducing the sleeping probability or the length of the sleeping period for these three strategies will reduce PAoI, and increasing the setup time variability for N-policy and multiple sleep strategy will also reduce PAoI.