Age of Processing: Age-Driven Status Sampling and Processing Offloading for Edge-Computing-Enabled Real-Time IoT Applications

• DOI: 10.1109/jiot.2021.3064055
• Publication Date: 2021-03-05T20:55:47Z
• Authors: Rui Li; Qian Ma; Jie Gong; Zhi Zhou; Xu Chen
• Keywords: Data Processing

The freshness of status information is great importance for time-critical Internet-of-Things (IoT) applications. A metric measuring the Age Information (AoI), which captures time elapsed from being generated at source node (e.g., a sensor) to latest update. However, in intelligent IoT applications such as video surveillance, revealed after some computation-intensive and time-consuming data processing operations, would affect freshness. In this article, we propose novel metric, Processing (AoP), quantify freshness, newest received processed since it generated. Compared with AoI, AoP further takes into account. Since an device has limited computation energy resources, can choose offload nearby edge server under constrained sampling frequency. We aim minimize <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">average</i> long-term process by jointly optimizing frequency offloading policy. first formulate online problem infinite-horizon Markov decision (CMDP) average reward criterion. then transform CMDP unconstrained (MDP) leveraging Lagrangian method, accordingly transformation framework original problem. Furthermore, integrate perturbation-based refinement mechanism achieving optimal policy Our investigation shows that AoP: 1) offloading: exploits good channel state 2) sampling: waiting presents threshold structure. Extensive numerical evaluations show proposed algorithm outperforms benchmarks, reduction up 30%.