A Decision Support System for Cyber Physical Systems under Disruptive Events: Smart Building Application

Abstract

Cyber-physical systems (CPS) are highly vulnerable to disruptions and disturbances during decision-making processes. An important challenge in CPS management is to anticipate, absorb, adapt to, and rapidly recover from unexpected critical situations. This study proposes a decision support system (DSS) for CPS management and planning under uncertainty. The DSS architecture is developed based on Markov Decision Process (MDP) and it reinforces the model-based CPS management system in making optimal decisions by taking into account a wide range of disruptive scenarios existing in CPS operation. In the event of a disruption, an initial decision is chosen randomly, and then it is optimized gradually over time by transiting to new states and finding decisions with higher reward values. Two factors are considered for determining immediate rewards in the support system: execution time and cost of the decision. The proposed decision support system is applied to tackle disruptions (including fire hazard, power outage, pipe burst, appliance malfunction, natural disaster, and compromised network) in a smart building case study. In contrast to the existing decision support systems presented in the literature, the proposed architecture is provided in a more integrated and generalizable manner that can be applied to various smart building decision-making applications during disruptive/critical events.