Modelling Approach For Cyber Physical Systems Based On SysML/MARTE/pCCSL

Cyber Physical Systems(CPS)are advanced embedded systems that combine digital computational systems with physical processes and focus more on the interaction and collaboration between computers and the physical environment.Since the concept was put forward,CPS have been highly concerned by the academic and industrial circles.CPS are mostly heterogeneous hybrid systems which are exposed to an open environment,and also combine continuous physical processes and discrete system behavior,therefore,CPS have the characteristics of heterogeneity,uncertainty,continuity and so on.At the same time,most CPS applications are safety-critical or strict in energy consumption,so a number of Non-Functional Properties(NFP)like time and energy consumption need to be met when functional ones have been guaranteed.Consequently,NFP of CPS are also to be studied.The modelling for such heterogeneous systems is actively researched.However,there still lacks a systematic approach to model the characteristics of CPS such as heterogeneity,stochastic behavior,continuous behavior and NFP,etc.In this paper,we propose a co-modelling approach that relies on SysML/MARTE/pCCSL to model different aspects of CPS from different point of views,including structure,behavior,clock constraints and NFP.Firstly,in order to capture the characteristics of CPS such as heterogeneity and continuous behavior,four SysML/MARTE meta-models are extended,including requirement diagram,block definition diagram,internal block diagram and state machine diagram.For the stochastic behavior of CPS,three operators are extended based on Clock Constrain Specification Language(CCSL)to obtain an extended language pCCSL.It can model the stochastic behavior of CPS and refine the model.Meanwhile,as a clock constraint language,pCCSL can specify the interaction and synchronization between different models and explicitly model the logical consistency among multi-view models.Secondly,in order to model CPS systematically,we propose a co-modelling approach that relies on SysML/MARTE/pCCSL to model different aspects of CPS from different point of views.The novelty of our approach lies in the use of logical clocks to drive and coordinate different models.Besides,based on model-driven method and standard visual modelling languages,we provide a co-modelling approach for the development of high-quality CPS.Thirdly,for the models refined by pCCSL,in order to explore the feasibility of alternative solutions and promote the refinement process,a mapping rule from MARTE/pCCSL to Stochastic Hybrid Automata(SHA)is proposed.By this way,we can use Statistical Model Checking(SMC)to varify and evaluate the models and help designers optimize system models.To implement the extended SysML/MARTE meta-model and pCCSL,we develope a GeMoC-based toolset.Finally,in order to demonstrate the feasibility of the proposed co-modelling approach and the extended meta-models,we model and refine the case of energy-aware building,then verify and evaluate the refined models.The experiment results prove that the comodelling approach is feasible and we can use SMC to verify the system.