Modeling Weighted Average Age of Information in Non-Symmetric Networks: Practical Scheduling Approaches

Abstract

In the ever-expanding environment of the Internet-of-Things (IoT), multiple sensors/- sources simultaneously send their updated information to the destination or sink. To ensure the updated data availability, a new concept Age of Information (AoI) has emerged as a key metric, representing the time elapsed since the last update from a specific sensor or source was received at the destination. A lower AoI indicates fresher information, which is crucial for maintaining data relevance. The minimum the AoI, the freshest will be the information, and vice versa. It becomes more difficult to minimize AoI when multiple users use multiple hops under non-symmetric network conditions to send their data to some remote destination. This thesis investigates average AoI in a non-symmetric multi-user multi-hop network. To manage the non-symmetry complexity imposed by the multi-user multi-hop scenario, only one channel is assumed for data transfer, from the source to the sink. Depending on the arrival time, each data packet is assigned a specific weight. These assigned weights serve as a reference in min- imization of the AoI. After assigning proper weights, next we schedule these weighted packets. For scheduling purposes, three scheduling policies, i.e., first source first serve (FSFS), earliest served first serve (ESFS), and single buffer with replacement (SBR), are implemented. The outputs of these policies are compared using the Cumulative Distribution Function (CDF), Sum of age of information, and age violation probability. The results demonstrate that the Earliest Served First Served policy consistently yields consistently yields minimum age of information in all considered aspects.