Towards Optimal Synchronization Scheduling in Internet of (Heterogeneous) Things

In this paper, the problem of optimal scheduling of network synchronization in the Internet of Things is studied. The optimal scheduling problem is transformed into a stochastic optimization problem by modeling clock skew and offset process as virtual queues. Synchronization is therefore achieved by controlling the virtual queue backlog sizes to satisfy constraints on synchronization errors. Accordingly, an optimal synchronization scheduling algorithm (OSSA) is designed to optimize network utility functions. Optimal policies are considered to minimize the energy cost and the age of information using OSSA, and it is proven that the performance can be controlled to be arbitrarily close to optimal. The performance of OSSA is compared against periodic and on-demand synchronization algorithms.