M Calder and A Miller. Using SPIN for feature interaction analysis - a case study. In M.B. Dwyer, ed. Proc. SPIN 2001, vol. 2057 LNCS, pp. 143-162, 2001. Springer-Verlag.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....LOTOS [7] is introduced. After features have been semi formally described with Use Case Maps, these features are formally defined with LOTOS, which then can be used for verification. Other authors also propose using formal description techniques for the purpose of interaction detection. In [8] and [9] temporal logic and model checking are suggested, in [10] a special heuristic is chosen, and in [11] as well as in [12] description logic and logic meta programming are applied. All of the above approaches require the modeler to explicitly and formally specify features. In contrast to this, our ....

M. Calder, A. Miller. "Using SPIN for Feature Interaction Analysis -- A Case Study" in M.B. Dwyer (Ed.) Model Checking Software 8th International SPIN Workshop. Toronto, Canada. 2001. pp. 143-- 162

....not isomorphic to the other processes. We therefore let q be p 1 . Note that the process q here should not be confused with a general environment process (see section 2) which need bear no relation to p. An example of this scenario is a communications service, with asynchronous communication [9]. Consider model checking a system of k 1 basic call processes acting concurrently together with a basic call process augmented with additional feature behaviour. Thus P k = q(x 1 )jjp(x 2 )jj : jjp(x k ) where (x 1 ; x 2 ; x k ) 1; 2; k) for some permutation . Suppose ....

M Calder and A Miller. Using SPIN for feature interaction analysis - a case study. In M.B. Dwyer, ed. Proc. SPIN 2001, vol. 2057 LNCS, pp. 143-162, 2001. Springer-Verlag.

....: t] is a temporal logic formula containing free variables indexed by 0; 1; t, where 0 t 2. The p i communicate peer to peer, asynchronously. In general, the p i are not isomorphic (because they have different sets of features enabled) We have demonstrated a number of such results [3] for a basic telecommunications service with features modelled in Promela. The properties are specified in linear temporal logic (LTL) and verified using the SPIN model checker [6] The [0; 1; t] express properties about feature behaviour (e.g. if process 1 has call forwarding to process ....

....but not the detailed description of the (parameterised) Promela process User, its variables and associated proposi tions. att(i; k) denotes the proposition a call is attempted from User[i] to User[j] which is itself defined in terms of the global variables associated with the User process (see [3]) Property 1 CFU Assume that User[j] has CFU User[k] If User[i] rings User[j] then a connection between i and k will be attempted before User[i] has handset on. LTL: p (rUq) p = dialed[i] j) User[i] calling) r = att(i; k) q = dev[i] on) 3 Feature validation and ....

[Article contains additional citation context not shown here]

M. Calder and A. Miller. Using SPIN for feature interaction analysis - a case study. In Proceedings of SPIN 2001.

....and 12. In Section 13 we automate the (dynamic) analysis and model generation and give results. In Section 14 we discuss the role of static and dynamic analysis. Conclusions and directions for further work are given in section 15. Some of the results contained in this paper have been presented in [5]. However here we include greater implementation detail and provide more background material relating to SPIN. In addition, our analysis concerns 8 features whereas in [5] only 6 were considered. 1.1 Related Work Model checking for feature interaction analysis has been investigated by others, ....

....and directions for further work are given in section 15. Some of the results contained in this paper have been presented in [5] However here we include greater implementation detail and provide more background material relating to SPIN. In addition, our analysis concerns 8 features whereas in [5] only 6 were considered. 1.1 Related Work Model checking for feature interaction analysis has been investigated by others, for example, using SMV [39] Caesar [41] COSPAN [17] and SPIN in [29] In the last, the Promela model is extracted mechanically from call processing software code; no ....

M Calder and A Miller. Using SPIN for feature interaction analysis - a case study. In [15], pages 143-162, 2001.

....communication protocols by exhaustive inspection of reachable composite system states in a nite state machine representation of the system. The SPIN model checker [Hol93] has been widely used in numerous veri cation case studies and industrial software controlled systems [LS97, CGM 97, CM01b, Hol99, HS99, HP00] In this paper we use SPIN to analyse the Tree Identi cation phase of the FireWire protocol. Model checking is by no means the push button technology that it is rumoured to be. It is crucial that the abstraction made of the system is both true to the original system and ....

....exploration of the entire state space. It is possible, for example, through ine cient abstraction, to completely fail to prove any properties of a system due to state space explosion. In this paper the authors apply their previous experience with the SPIN model checker (see for example [CM01b] to model and analyse the FireWire protocol. By applying the optimisation and model generation techniques described in [CM01b] we are able to create a model with a tractable state space and perform analysis upon it so allowing for a fair comparison with the other veri cation methods discussed ....

[Article contains additional citation context not shown here]

M Calder and A Miller. Using SPIN for feature interaction analysis - a case study. In [Dwy01], pages 143-162, 2001.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC