“…Some of (centralized) deterministic algorithms, e.g., [18], rely on (commercial) solvers which may not be scalable to very large systems to meet the requirement of realtime control. b) Controlling Devices with Dynamics: Controllable devices with dynamics are usually handled in two ways: control based on heuristic or engineering intuition, see, e.g., [19] that controls TCLs based on temperature status; and optimizationbased control that can integrate specific objective functions and constraints, see, e.g., [20] that considers device dynamics but does not involve discrete variables, [21] that considers device dynamics and discrete decision variables but employs commercial solvers, [22], [23] that solve OPF over various devices with dynamics but consider equality constraints only, and [24], [25] that formulate mixed-integer linear programs which cannot be applied to solving more general problems with convex cost functions. c) Market-Based/Demand-Response Design: Existing literature on market-based and demand-response problem formulation mostly focuses on demand/supply balancing, without considering network structure, see, e.g., [3]- [6], [20], [21], [26].…”