A hybrid approach of semantic modeling and co-simulation for a better consideration of physical phenomena in a smart building
Abstract
Physical phenomena occurring in a complex and heterogeneous connected cyber-physical system (e.g. smart building) are poorly taken into account in current IoT (Internet of Things) applications.
These physical phenomena are often interdependent and interact according to physical laws that are completely out of control of current IoT systems. For example, thermal transfers from one part of the environment to another (e.g. from the living room to the hallway...) are difficult to formulate. According to the concept of digital twins, co-simulation approaches based on observations of connected objects allow to anticipate the temporal evolution of the cyber-physical system and, ultimately, choose the actions to be performed by these same objects to optimize the control of the system. The main objective of this thesis is to design a hybrid approach of semantic modeling and co-simulation for a better consideration of physical phenomena in smart buildings. Complex cyber-physical systems such as smart buildings are composed of several interacting heterogeneous subsystems. This makes the global dynamics of the cyber-physical system difficult to model. This heterogeneity implies the modeling of each subsystem with a specific tool. These models are then coupled in a co-simulation to reproduce the behavior of the global system.
In this context, we will study how dynamic system models describing these physical phenomena can be integrated within the IoT systems and how these models are exchanged and co-simulated in a distributed manner within the IoT network of a cyber-physical system.
From the general objective, three issues arise: 1) how to ensure interoperability between the simulations and the IoT system? 2) how to embed the simulations in the constrained objects of an IoT system? and 3) which co-simulation approach to manage distributed simulations?
The first sub-objective is to integrate physical phenomena into the IoT system, which would improve the accuracy of the reasoning system and energy efficiency. The difficulty of this step lies in the interoperability between the dynamic model and the IoT system.
The second sub-objective is to embed the simulation in the constrained IoT devices, which would reduce the response time and ensure the minimum system operation in case of communication problems. The challenge in this part is to figure out a way to embed these simulations on constrained objects. In fact, a simulation is relatively greedy in terms of resources, whether in terms of memory, computing power, or energy. The issue would be how to make the simulation light (less greedy in resources) and adapted to the architecture of the constrained object.
The third objective is to co-simulate distributed simulations using semantic web technologies which would improve the interoperability of the IoT system. The question is how to handle the data exchange between interdependent simulations using semantic web technologies? In the poster attached below, I'll present the context and the three objectives we mentioned above. I present a use case and the first approach we are developing to address the first objective, which is to integrate physical phenomena into the IoT system using Semantic Web technologies.