1 Simulation-based Parameter Optimization Framework for Large-Scale Hybrid Smart Grid Communications Systems Design Adarsh Hasandka 1 *, Jianhua Zhang 2 , S M Shafiul Alam 2 , Anthony R. Florita 2 and Bri-Mathias Hodge 1,2 1 Department of Electrical, Computer and Energy Engineering, University of Colorado, Boulder, CO 80309, USA 2 National Renewable Energy Laboratory (NREL) Golden, CO 80401, USA Email: Adarsh.Hasandka@colorado.edu, {Jianhua.Zhang, SMShafiul.Alam, Anthony.Florita, Bri.Mathias.Hodge}@nrel.gov Abstract—The design of reliable, dynamic, fault-tolerant hy- brid smart grid communication networks is a challenge to achieve for autonomous power grids. A simulation-based param- eter optimization framework is proposed to tune parameters of hybrid communication technologies to achieve the optimal network performance. It consists of three main components: a parallel executor used to speedup a list of simulations; a sampler running simulations with all possible parameter sets for the input parameter variables by using the parallel executor at each generation; and a hybrid stochastic optimization algorithm for tuning configurable parameters of hybrid designs and applica- tion parameter variables. The proposed hybrid metaheuristic optimization algorithm combines an evolutionary algorithm with a gradient method to quickly achieve an approximate globally optimum solution. Three optimization test functions are employed to train the adjustable parameters of the hybrid algorithm. Re- sults show that the proposed parameter optimization framework helps the designer to choose the right hybrid architecture with an optimal parameter set for a large-scale broadband PLC-WiMAX hybrid smart grid communication network. I. I NTRODUCTION The power grids worldwide are evolving into smart grids by adding intelligence to the operation and control of the system[1], [2], [3]. As a result, it becomes increasingly important to explore the communication capabilities of different types of smart grid topologies [4], [5]. The envisioned smart grid communication network for distributed applications broadly consists of Home Area Networks (HAN), Neighborhood Area Networks (NAN), and a Wide Area Network (WAN), and it is expected that a variety of communication technologies will be utilized in the hybrid communications systems infrastructure [6]. The design of hybrid communication networks are not straightforward, because the different technologies used in different sub-network have a large number of configurable parameters which increases the amount of experimental (or simulation) tests necessary for their evaluation. The non- deterministic nature of the environment is another factor which makes network design difficult. The hybrid smart grid communication network must be fault tolerant and adaptive because of the dynamic network topology caused by dynamic power grid topology and changing objectives of different smart grid applications. The design of a hybrid smart grid communication network requires a simulation-based optimization method to tune the configuration parameters of communication technologies and parameters of smart grid applications. A simulation-based parameter optimization framework is proposed in this paper to help the designer choose the right hybrid architecture with an optimal parameter set. This scalable and extendable framework may accept different communication technologies on top of different topologies, and identify the optimal configurable parameters for each related communications model and application parameters for that hybrid design. The novel contribution of this work is the simulation-based parameter optimization framework with features of parallel computing and using a hybrid evolutionary search algorithm. The proposed design provides a simulation-based optimization tool than can help designers identify the optimal parameter set for a selected hybrid communication configuration. Although the hybrid evolutionary search agorithm has previously been investigated in [7] and [8], this algorithm is utilized in this paper to develop a new tool that performs network parameter optimization. While there has been similar previous work done to simulate hybrid communication networks as in [9] and [10], the proposed algorithm has been designed specifically to be used with NS3. NS3 is an open-source discrete-event network simulator capable of simulating many important aspects of communication technologies: such as propagation model, spectral model, payload modulation coding scheme, and service flow type [11]. Using the simulator, the optimization algorithm tunes all the input parameters, at both the application and architectural level, to provide an optimum set within the required QoS metrics. The large parameter space and the simulation-based genetic algorithm impose a heavy computational load. It is beneficial to parallelize execution of these computationally intensive simulations and thus speed- up the performance of the simulation-based optimization algorithm. Through combining a gradient-based algorithm and a genetic algorithm, the hybrid evolutionary gradient algorithm is proposed as a new parameter identification algorithm. The primary application of this framework is thus the optimization of network configuration parameters and application parameters through extensive hybrid communication system simulations. The proposed solution provides a way to design and optimize hybrid smart grid communication systems in a highly non- deterministic environment for a large number of cooperating intelligent power grid devices.