27.03.2014 Views

SEKE 2012 Proceedings - Knowledge Systems Institute

SEKE 2012 Proceedings - Knowledge Systems Institute

SEKE 2012 Proceedings - Knowledge Systems Institute

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

A HybridUML and QdL Based Verification Method<br />

for CPS Self-Adaptability<br />

Jiakai Li, Bixin Li, Qiaoqiao Chen, Min Zhu, Shunhui Ji, Xiaoxiang Zhai<br />

School of Computer Science and Engineering, Southeast University, Nanjing, China<br />

{jiakai_li, bx.li}@seu.edu.cn<br />

Abstract—CPS (Cyber-Physical <strong>Systems</strong>) are physical and<br />

engineered systems featuring a tight combination of computation<br />

and physical processes by communication networks. CPS are<br />

mainly applied in some critical domains, so it is very essential to<br />

ensure the correctness of CPS. Formal verification has been<br />

successfully applied in the correctness verification of CPS;<br />

however, the high theoretical level of formal modeling techniques<br />

of formal verification and the lack of visualization of formal<br />

models make it difficult to integrate formal verification with<br />

enterprise standard system development process. In this paper,<br />

we model CPS by HybridUML, an extension of UML, and then<br />

transform HybridUML model into the input language of<br />

theorem prover KeYmaera-QHP(Quantified Hybrid Program),<br />

and finally verify the QHP code with KeYmaera.<br />

Keywords-CPS; Formal Verification; HybridUML; QHP;<br />

MetaModel; Model Transformation<br />

I. INTRODUCTION<br />

CPS, in a broad sense, are controllable, trustable and<br />

extensible networked physical device systems that deeply<br />

integrate computation, communication and control capabilities<br />

based on environment perception [1] . CPS are mainly applied in<br />

areas having a high demand on performance, so to ensure the<br />

correctness of CPS is extremely important.<br />

CPS evolve over time with interacting discrete and<br />

continuous dynamics, which accords with the definition of<br />

hybrid systems, so the verification theory of hybrid systems<br />

can be referenced in CPS verification. Although many present<br />

model checking [2,3,4,5] and theorem proving methods [6,7,8] have<br />

been proposed for Hybrid <strong>Systems</strong> verification, these<br />

methods have great deficiencies in supporting the verification<br />

of s ystems with distributed characteristics. To solve this,<br />

André Platzer proposed a verification technique based on QdL<br />

(Quantified Differential Dynamic Logic) [9,10,11] . In order to<br />

achieve automatic verification, André designed a theorem<br />

prover- KeYmaera. However, the high theoretical level of<br />

formal modeling method in QdL and the lack of visualization<br />

of QHP make it difficult to integrate QdL based formal<br />

verification with enterprise system development processes in<br />

which the Unified Modeling Language (UML) has been<br />

This work is supported partially by National Natural Science Foundation of<br />

China under Grant No. 60973149, partially by the Open Funds of State Key<br />

Laboratory of Computer Science of Chinese Academy of Sciences under Grant<br />

No.SYSKF1110, partially by Doctoral Fund of Ministry of Education of China<br />

under Grant No. 20100092110022, and partially by the College Industrialization<br />

Project of Jiangsu Province under Grant No.JHB2011-3.<br />

accepted as the de facto standard modeling language.<br />

However, UML lacks precise formal semantics, making it<br />

hard to directly verify UML models formally. The paper first<br />

models CPS by HybridUML, and then translates the model<br />

into the input model of KeYmaera [12] , and finally performs<br />

formal verification with KeYmaera to ensure the correctness<br />

of CPS.<br />

The interconnection topology and the number of active<br />

members of CPS network, especially those composed of<br />

mobile devices such as the Distributed Air Traffic Collision<br />

Avoidance System (DATC), tend to change dynamically,<br />

making how to ensure critical properties hold dynamically<br />

really challenging. The characteristic that CPS adjust their<br />

behaviors flexibly according to the network structural and<br />

dimensional dynamics is called self-adaptability. Verifying<br />

self-adaptability, Hierarchical Hybrid StateMachine-Mode is<br />

chosen as the CPS modeling view. First, the hierarchical<br />

feature of the top-level Mode model is eliminated according to<br />

execution semantics and a FlatMode model is gotten. Second,<br />

transform the FlatMode model to a QHP model by executing<br />

corresponding ATL transformation rules. Third, template rules<br />

in a customized template language will be called to transform<br />

the QHP model to QHP code, and then specify the CPS selfadaptability<br />

property with a QdL formula . Finally, verify<br />

property formula automatically with KeYmaera.<br />

The paper is organized as follows. Section introduces<br />

HybridUML based modeling method. Section describes the<br />

translation from HybridUML model to QHP code. Section <br />

illustrates the application of our verification architecture by<br />

a case study of a DATC. Section summarizes the paper and<br />

makes a discussion about the future work.<br />

II.<br />

CPS MODELING BASED ON HYBRIDUML<br />

A. Basic modeling elements of HybridUML<br />

1) Agent<br />

In HybridUML, Agent is a stereotype for Class in UML.<br />

As the basic building block for describing the static system<br />

architecture, an Agent can own an internal structure consisting<br />

of other Agent instances to support hierarchical models [13] .<br />

2) Hierarchical Hybrid StateMachine –Mode<br />

In HybridUML, Mode is an extension of UML<br />

StateMachine and is used for describing Agents′ behavior [13] .<br />

As a Hierarchical Hybrid StateMachine, a Mo de can contain<br />

sub-modes and transitions inside its region. When a Mode is<br />

239

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!