On Pursuit of Sleep-scheduling for Sensor Network in IIoT: A Non-linear AoI Optimization Perspective

Freshness of data holds significant importance in the Industrial Internet of Things (IIoT), giving rise to Age of Information (AoI). In practical scenarios, data freshness experiences an accelerated decline. Yet linear AoI fails to depict the variation of information freshness. Therefore, we present a non-linear AoI characterized by a variable growth rate. Furthermore, with prolonged operational duration, failure rate of sensors rises, resulting in an elevated probability of transmission outages and, consequently, an increase in AoI. Recognizing this correlation, we introduce a sleep mechanism to mitigate AoI while saving energy. Although sleep mechanism helps to AoI reduction and energy conservation, it also leads to a low utilization rate of sensors. Considering this, we formulate an non-linear AoI minimization problem to optimize sleep decision under energy and utilization rate constraints. Then, a proximal policy optimization based sleep-scheduling algorithm (PSS) is proposed to make sleep decisions. Numerical results demonstrate the superiority of our proposed non-linear AoI as well as our PSS schemes.