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.

protocol of DATS we used is properly modified from [11].<br />

The differential equation set representing the flight dynamics<br />

of an aircraft i of type A is shown in formula F w(i) (i) in which<br />

x 1 (i) and x 2 (i) denote the shift of i in the x-axis and y-axis<br />

respectively, d 1 (i) and d 2 (i) represent the velocity of aircraft i<br />

in the x-axis and y-axis respectively, and w(i) represents the<br />

angular velocity of i.<br />

x 1 (i)′=d 1 (i)∧x 2 (i)′=d 2 (i)∧d 1 (i)′=-w(i)d 2 (i)∧d 2 (i)′=w(i)d(i) (F w(i) (i))<br />

Two aircraft i and j satisfy safety separation property if<br />

and only if the following formula holds:<br />

S(i, j)≡( x 1 (i)- x 1 (j)) 2 +( x 2 (i)- x 2 (j)) 2 ≥P 2 ∨ i=j<br />

At the beginning, all aircraft are fa r apart from one<br />

another, and they fly freely with angular velocity w(i).<br />

However, when the distance between aircraft is no larger than<br />

∂ that is positively proportional to P and the radius of the<br />

roundabout circle, all involved aircraft will come to an<br />

agreement on a c ommon angular speed w and a roundabout<br />

circle center c(c 1 ,c 2 ), then through phase Entry, each aircraft<br />

reaches a tangential location around center c. During Circle<br />

phase, each aircraft flies along the roundabout circle<br />

tangentially with the agreed common angular speed w. When<br />

a new aircraft dynamically appears in the collision avoidance<br />

zone, all aircraft again negotiate a new common angular speed<br />

w and roundabout circle center c, and then each aircraft goes<br />

through phase Entry and Circle again. Once the maneuver<br />

finishes, each aircraft will continue to fly in the original<br />

direction and enter phase free again. Adaptive collision<br />

avoidance protocol is illustrated in Figure 2 in the form of a<br />

Mode model. After the transformation of the model, final<br />

QHP code shown in Figure 4 is gotten.<br />

B. Self-Adaptability specification and verification<br />

In this case, we focus on the verification of adaptive<br />

collision avoidance property, i.e., when a new member enters<br />

the collision avoidance zone of existing aircrafts, the collision<br />

avoidance property still holds. Let φ≡∀i,j:A S(i, j), and the<br />

property verified can be specified in the following formula:<br />

φ→[CAMP]φ. The formula expresses that φ will always hold<br />

after evolving along CAMP for an arbitrary length of time if φ<br />

holds in the initial condition. In QdL, property verification is<br />

accomplished in the form of QdL proof calculus for which<br />

Sequent Calculus, which works by QdL proof rules, is<br />

adopted as the basic proof system [9] . During the proof, the<br />

proven property formula is placed at the bottom of the whole<br />

process, and the proof calculus is carried out from the bottom<br />

up. If the property finally holds, then the proof process ends<br />

up with a star symbol, otherwise the precondition that is<br />

required for the formula to hold is given. The a bbreviated<br />

proof process that KeYmaera executes inside is shown in<br />

Figure 5. That all branches end successfully means the<br />

verified self-adaptability property holds.<br />

true true true true<br />

DR ' ax<br />

[?] ax<br />

φ [ M1] φ φ [ free] φ φ [? true] φ φ [( newP; M1)]<br />

φ<br />

[] gen<br />

[ ]<br />

φ [ α] φ φ [ β]<br />

φ<br />

[;][] gen<br />

φ [ α; β]<br />

φ<br />

→ rind ,<br />

φ → [ M *] φ<br />

Figure 5. Property proof process<br />

V. CONCLUSION<br />

The paper proposes a QdL based verification architecture<br />

for CPS self-adaptability property. First, CPS are modeled by<br />

HybridUML. Second, the model expressed by HybridUML is<br />

transformed into the input language of KeYmaera-QHP. Third,<br />

combined with the QHP code, the property to be proven is<br />

specified in the form of a QdL formula. Finally, the QdL<br />

formula is verified automatically using KeYmaera. In order to<br />

achieve the automatic translation from HybridUML Mode to<br />

QHP code and automatic verification by effective integration<br />

with KeYmaera, our future work will focus on t he<br />

implementation of a model transformation tool.<br />

ACKNOWLEDGMENT<br />

The authors thank all the teachers and students who<br />

provide support for our work. Particularly, we thank Qi<br />

ShanShan and Liu CuiCui for their substantial work during<br />

paper correction.<br />

REFERENCES<br />

[1] He JF. Cyber-physical <strong>Systems</strong>. Communication of the CCF,<br />

2010,6(1):25-29.<br />

[2] Clarke EM, Emerson EA, Sifakis J. Model checking: algorithmic<br />

verification and debugging. Communications of The ACM, 2009,<br />

52(11):74–84 .<br />

[3] Alur R, Courcoubetis C, Halbwachs N, et al. The algorithmic analysis of<br />

hybrid systems. Theoretical Computer Science, 1995,138(1): 3–34.<br />

[4] Henzinger TA. The theory of hybrid automata. Proc. of the 11th Annual<br />

IEEE Symposium on Logic in Computer Science. 1996. 278-292.<br />

[5] Clarke EM, Zuliani P. Statistical Model Checking for Cyber-Physical<br />

<strong>Systems</strong>. T. Bultan and P.-A. Hsiung (Eds.): ATVA 2011, LNCS 6996,<br />

2011,1–12.<br />

[6] Manna Z, S ipma H. Deductive verification of hybrid systems using<br />

STeP. Proc. of Hybrid <strong>Systems</strong>: Computation and Control, First<br />

International Workshop, HSCC 98, Berkeley, California, USA.<br />

California: Springer. 1998. 305–318<br />

[7] Van Beek DA, Man KL, Reniers MA, et al. Syntax and consistent<br />

equation semantics of hybrid Chi. Journal of Logic and Algebraic<br />

Programming, 2006, 68(1-2):129–210.<br />

[8] Zhou CC, Ravn AP, Hansen MR. An extended duration calculus for<br />

hybrid real-time systems. Hybrid <strong>Systems</strong>, 1993, 7: 36–59.<br />

[9] Platzer A. Quantified Differential Dynamic Logic for Distributed<br />

Hybrid <strong>Systems</strong>. In: A.Dawar, H.Veith, eds. Proc. of CSL 2010, LNCS<br />

6247. Heidelberg: Springer-Verlag, 2010. 469-483.<br />

[10] Platzer A. Quantified differential dynamic logic for distributed hybrid<br />

systems. Technical Report, CMU-CS-10-126, SCS, Carnegie Mellon<br />

University, 2010.<br />

[11] Platzer A. Quantified Differential Invariants. In: Emilio Frazzoli, Radu<br />

Grosu, eds. Proc. of the 14th ACM International Conference on Hybrid<br />

<strong>Systems</strong>: Computation and Control, HSCC 2011, Chicago, USA, April<br />

12-14, 2011, Chicago: ACM, 2011. 63-72.<br />

[12] Platzer A, Quesel JD. KeYmaera: A hybrid theorem prover for hybrid<br />

systems. In: Alessandro Armando, Peter Baumgartner, et.al, eds. Proc.<br />

of IJCAR 2008, LNCS 5195, Heidelberg: Springer, 2008.171-178.<br />

[13] Berkenkötter K, Bisanz S, Hannemann U, et al. The HybridUML profile<br />

for UML 2.0. International Journal on Software Tools for Technology,<br />

2006, 8(2):167–176.<br />

[14] Alur R, Grosu R, Lee I, et al. Compositional modeling and refinement<br />

for hierarchical hybrid systems. In <strong>Proceedings</strong> of J. Log. Algebr.<br />

Program.. 2006, 105-128.<br />

[15] Jouault F, Allilaire F. ATL: A model transformation tool. Science of<br />

Computer Programming, 2008, 72(1-2):31-39.<br />

242

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

Saved successfully!

Ooh no, something went wrong!