Modelling and simulation of non-functional properties in cyber-physical systems

Cyber Physical Systems (CPS) have emerged as one of the most important academic and industrial research topics. These systems bridge the cyber world of computing and communications with the physical world. This intimate coupling between the cyber and physical manifests itself from the nano-world to large-scale systems. The CPS will transform how humans interact with and control the physical world around us. As a very promising research field, there are a variety of questions need to be solved, at different layers of the architecture and for different aspects of systems design, to trigger and to ease the integration of the physical and cyber worlds. Therefore, the next challenge facing the EDA community is to develop methods and also tools for CPS design. However, CPS as a tightly integrated system with computation and physical components, currently available EDA methods and tools are not equipped to handle such integrated modelling and simulation. To address this issue, there is a demand for developing appropriate EDA methods and tools to satisfy the need for an integrated modelling of the dynamics of the physical systems and the computational platforms. Because CPS include heterogeneous components, even the simulation of their functionality represents already a significant challenge, since these components imply different time scales and accuracies, and interactions among different domains. However, the assessment of functionality is not the only dimension to be considered. Since CPS combine heterogeneous domains, other metrics must be considered to ensure their correct operations, such as power consumption, thermal behaviour or reliability. Modelling and monitoring of these properties in isolation is not a new problem. Power, thermal and reliability analysis have been studied since two decades in traditional EDA field, unfortunately, no current design methodology can simultaneously master all non-functional aspects of CPS. This dissertation aims at bridging this gap focusing on integration-aware solutions to different non-functional aspects of CPS by proposing a novel multi-layer, bus-centric single modelling and simulation framework. Multi-layer because it is structured hierarchically, with each property corresponding to a simulation layer; Bus-centric in that each property is simulated by adopting a specific virtual bus, which conveys and elaborates property information, used to derive property-specific status of the overall system. This dissertation focuses on three main non-functional properties: power, temperature and reliability, in the form of aging. Concerning power property, firstly, a unified model of a power source that is applicable to any power scale, and that can be derived solely from data contained in the specification is proposed; then we propose a novel methodology that bring the traditional analysis of Energy Storage Devices (ESDs) in the time domain move to frequency domain, a circuit equivalent battery model account for load variations is also proposed. Based on the proposed novel battery model, we propose an optimal battery-aware scheduling policy for two kinds of CPS, namely, a multi-sensor IoT device and unmanned aerial vehicle delivery system. Moreover, since the ESDs of a CPS also need recharging when it finishes discharging phase, and the charging protocol strongly affects the performance of ESDs, we introduce an optimal charging protocol to reduce the battery aging effects while keep high Quality of Service (QoS) of the battery powered devices. In terms of temperature, since the existing methods rely on the co-simulation with an external thermal simulator, we devise one built-in thermal simulator in our proposed simulation framework. For achieving concurrent co-simulation with others non-functional properties in our proposed framework, we implement the standard RC equivalent circuit thermal simulator into a SystemC-AMS platform, we get comparable thermal simulation results to the existing simulators, yet with much better performance. With regard to aging, in order to keep the same simulation speed with other non-functional properties, the existing transistor-level and gate-level methods for investigating the aging, specifically, the aging due to Negative-bias temperature instability (NBTI) of the microprocessor are inadequate, therefore, we propose a novel empirical system-level NBTI aging macro-model of microprocessor, we also propose one methodology that conducts characterization on different architectures and with various synthesis constraints allows to yield lower and upper bounds of NBTI aging of any embedded cores without knowing its net-list.