The event-driven paradigm for control, communication and optimization
Christos G. Cassandras · Journal of Control and Decision · 2014
The event-driven paradigm offers an alternative to the time-driven paradigm for modelling, sampling, estimation, control and optimization. This has come about largely as a consequence of systems being increasingly networked, wireless and consisting of distributed communicating components. The key idea is that control actions need not be dictated by time steps taken by a “clock”; rather, an action should be triggered by an “event” which may be a well-defined condition on the system state, including the possibility of a simple time step, or a random state transition. We provide an overview of recent developments in event-driven approaches and focus on two areas to illustrate their value. First, in distributed systems, we describe how event-driven, rather than synchronous, communication can guarantee convergence in cooperative distributed optimization while provably maintaining optimality. Second, in hybrid systems where events naturally decompose state trajectories into different discrete states (modes), we review the theory of infinitesimal perturbation analysis (IPA) which offers an event-driven “IPA calculus” for evaluating (or estimating in the case of stochastic systems) gradients of performance metrics, thus facilitating the solution of a large class of control and optimization problems.