Joint Wireless and Computational Resources Allocation for Optimizing the Age of Correlated Information in Fog Computing Networks

Age of Information (AoI) has gradually become a key performance indicator in many practical Internet of Things (IoT) systems. With the supplement of fog computing networks, the age of correlated information (AoCI) is extended from AoI to measure the information timeliness for multi-device aggravating applications. To capture all asynchronous devices' historical information, packet number-based AoCI (PN-AoCI) is introduced. Furthermore, the newest packet number-based AoCI (NPN-AoCI) is built to acquire information, where the newest information is more valuable than the old. Optimizing the peak AoCI under the above two scenarios is challenging due to the coupling of the ambiguous objective function and the high-dimensional feasible region. The approximate dynamic programming (ADP) concept is introduced to schedule wireless and computational resources jointly. By dividing the long-term optimization problem into short-term forms, the ADP can optimize both the PN-AoCI and NPN-AoCI in a unified manner. Then, the successive convex approximation (SCA) is used to get the myopic scheduling policy. Disciplined quasi-convex programming (DQCP) is used to build the noncontinuous look-ahead scheduling policy. Finally, the simulation results depict that the policies' performance is practical, and the packet generation policy is also discussed.