There is a lot of confusion and ambiguity regarding the quantification of the Quality of Service (QoS) of a system, especially for cyber-physical systems (CPS) involved in automating or controlling the operations in built environments and critical urban infrastructures, such as office buildings, factories, transportation systems, smart cities, etc. In these cases, the QoS, as experienced by human users, depends on the context in which they (i.e., humans) interact with these systems. Traditionally, the QoS of a CPS has been defined in terms of absolute metrics. Such measures are unable to take into account the variations in performance due to contextual factors arising out of different kinds of human interactions. Further, the QoS of a CPS has typically been evaluated by comparing the performance of the actual, fully realized system with the given QoS constraints only after the actual system has been completely developed. In the case of faults in the design exposed by observed deviations from the QoS constraints due to unpredicted variations in the contextual factors, the system needs to be re-designed and re-developed from scratch. Due to the above-mentioned reason, the validation approach associated with the traditional QoS makes the design of CPS systems prohibitively expensive, impractical, as well as infeasible in numerous application areas, such as civil and engineering works, since it may not be possible to modify the system once developed beyond a certain extent. To that end, we propose a context-aware definition of QoS of a CPS which facilitates the design of robust systems as elaborated below. In this paper, we define QoS as a function of contextual factors. A CPS designed according to our QoS specifications would always satisfy the QoS irrespective of any possible changes in contextual factors resulting from many different human interactions that may occur during operation of the system. We also present QACDes - a novel framework that provides a formal mechanism for validating the design of a CPS with respect to the specified QoS constraints at the design phase as well as after the realization of the actual system. QACDes can validate any given CPS, irrespective of its application domain, against a QoS guarantee: (A) as early as even before the design phase by comparing the proposed model with a baseline model, or (B) after the realization of the actual system based on logs collected from running the actual system. We consider a lighting control system that manages the light switches - switching it on/off depending on contextual factors, such as the presence of occupants and time of the day. Using the lighting control system in a building as a use case, we analyze and demonstrate the effectiveness of our QoS definition as well as the QACDes framework against the performance metric measured in an actual fully-realized CPS.

The increasing demand for demand-side flexibility is transforming residential buildings into active integrated energy systems (IESs), where distributed energy resources can be coordinated under dynamic electricity pricing while maintaining occupants’ thermal comfort. However, existing studies either simplify buildings as single thermal zones or adopt multi-zone formulations that still apply largely uniform comfort constraints and control priorities across zones. Such approaches overlook differences in occupant thermal vulnerability, potentially exposing vulnerable occupants to prolonged thermal discomfort and increasing the likelihood of control overrides, thereby reducing the flexibility that can be reliably delivered. This study develops a vulnerability-aware model predictive control (VA-MPC) strategy for residential IES management. A vulnerability-aware thermal comfort framework is proposed through spatial, temperature, and temporal constraints, enabling protection to follow the vulnerable occupant across zones, enforce a minimum protective temperature, and limit continuous exposure outside the preferred thermal region while retaining coordinated energy scheduling. Compared with economic MPC and economic MPC with a whole-building hard temperature constraint, VA-MPC reduces the vulnerable occupant’s thermal preference deviation by 80.5% and 54.2%, respectively, while keeping the longest uninterrupted exposure outside the preferred thermal region within the prescribed limit. This protection requires only a 1.8% increase in electricity cost relative to economic MPC, while remaining 3.2% and 9.7% cheaper than hard-constraint MPC and rule-based control, respectively. Ablation and control-behaviour analyses further show that protection is achieved through targeted spatial and temporal redistribution of heating. These results demonstrate that occupant thermal vulnerability can be explicitly incorporated into residential demand-side flexibility while preserving economic performance.

Virtual reality (VR) offers promise as a tool for building performance simulations, especially when considering human-building interactions in buildings or spaces still under design. However, the absence of standardized data protocols impedes the consistent sharing of VR-related experiments and findings. This makes advancing VR experimentation as a reliable method for studying human-building dynamics challenging. The authors introduced the Virtual Human-Building Interaction Experimentation Ontology (VHBIEO) to address the challenge. VHBIEO seeks to standardize experimentation details as a domain-specific ontology, enhancing their interoperability. It includes essential experimentation concepts and employs semantic web technologies to ensure machine readability. Moreover, it integrates an application view (APV) to tailor details to specific experiments. Using VHBIEO-based metadata, this paper presents a case study aiming to standardize experiments that validate thermal sensations in immersive virtual environments (IVE), encompassing experimental protocol, variables, design, and data gathering. By exploring the main characteristics of VHBIEO-based metadata, the authors discuss its potential to improve the reliability of human-building interaction research