H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional performance modelling with the TIPPtool. Performance Evaluation, 39(1-4): 5--35, 2000.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... Conclusions and Future Work We have introduced PRISM, a tool to build and analyse probabilistic systems which supports three types of models (DTMCs, MDPs and CTMCs) and two probabilistic logics (PCTL and CSL) Several DTMC and CTMC analysis tools are available, for example MARCA [29] and TIPPtool [19], which do not allow logic speci cations and instead support steady state and transient analysis. Of the two probabilistic model checking tools that we are aware of, ProbVerus [4] only supports DTMCs and a subset of PCTL, whereas E MC 2 [20] only supports the model checking of CTMCs using CSL ....

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional performance modelling with the TIPPtool. Perf. Eval., 39(1-4), 2000.

....for SPA, are discussed, with a breakdown of their strengths and weaknesses. The problems each has suggest features that an alternative specification method might possess. Such an alternative, the PEPA Reward language, is described in Section 3.2. 40 3.1. 1 Using Regular Expressions In [42], Hermanns and Mertsiotakis describe the reward specification mechanism used in their TIPPtool, an application for analysing SPA performance models. Given an SPA model, the CTMC is formed by a standard construction from the labelled transition system, itself formed using operational inference ....

....models is uneasy. 42 3.1. 2 Instrumenting SPA Models with Rewards An alternative reward specification technique for SPA is advocated by Bernardo [5] The method uses the Markovian process algebra EMPA; however variants of this method have been described for both PEPA, in [44] and TIPP, in [42]. The idea is not to use a separate formalism for specifying rewards, but rather to extend the language of the process algebra itself. This is done by adding a parameter to activities, denoting a reward value; for instance, #, r) becomes (#, r , #) meaning that any model state enabling this ....

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional performance modelling with TIPPtool. Performance Evaluation, 39:5--35, 2000.

No context found.

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional performance modelling with the TIPPtool. Performance Evaluation, 39(1-4): 5--35, 2000.

No context found.

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional performance modelling with the TIPPtool. Performance Evaluation, 39(1-4):5--35, January 2000. 294, 295, 303

No context found.

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, M. Siegle, Compositional performance modelling with the TIPPtool, Performance Evaluation 39 (1-4) (2000) 5--35.

No context found.

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, M. Siegle, Compositional performance modelling with the TIPPTOOL, in: R. Puignajer, N.N. Savio, B. Serra (Eds.), Computer Performance Evaluation, Lecture Notes in Computer Science, vol. 1469, Springer, Berlin, 1998, pp. 51-- 63.

No context found.

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis and M. Siegle. Compositional performance modelling with the TIPPtool. Performance Evaluation 39(1-4):5--35, 2000.

No context found.

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional performance modelling with the TIPPtool. Performance Evaluation, 39(1-4):5--35, January 2000.

No context found.

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional Performance Modelling with the TIPPtool. In R. Puigjaner, N. Savino, and B. Serra, editors, 10th International Conference on Modelling Techniques and Tools for Computer Performance Evaluation (TOOLS '98), pages 51--62, Palma de Mallorca, September 1998. Springer, LNCS 1469.

....53 489 4661. level, high level specification methods have been developed, most notably those based on stochastic Petri nets, stochastic process algebras, and stochastic activity networks. With appropriate tools supporting these specification methods, such as, for instance, provided by TIPPtool [36], the PEPA workbench [23] GreatSPN [13] UltraSAN [56] or SPNP [14] it is relatively comfortable to specify performance models of which the associated CTMCs have millions of states. In combination with state of the art numerical means to solve the resulting linear system of equations (for ....

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional performance modelling with the TIPPtool. Performance Evaluation, 39(1-4): 5--35, 2000.

....the states of the Markov chain with sets of atomic propositions, the matrix R constitutes the interface between the high level formalism at hand and the model checker. Currently, the tool accepts DTMCs and CTMCs represented in a format generated by the stochastic process algebra tool TIPPtool [13], but the tool is designed in such a way that it can easily bridge to various other input formats. The stochastic Petri net tool DaNAMiCS [4] has recently been extended to generate input for . Tool architecture The tool has been written entirely in Java (version 1.2) in order to provide ....

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis and M. Siegle. Compositional performance modelling with the TIPPtool. Performance Evaluation, 39(1-4): 5--35, 2000.

....the size of the MTBDD. However, similar phenomena can be observed when performing symbolic elimination of vanishing states or symbolic bisimulation: The size of the symbolic representation grows The elementary transition systems were generated by the stochastic process algebra tool TIPPtool [54] and then encoded as individual MTBDDs. Since the considered version of the polling model does not contain immediate transitions, a single MTBDD (representing Markovian transitions) is sufficient. although the underlying transition system is reduced, i.e. fewer states and transitions are ....

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional performance modelling with the TIPPtool. Performance Evaluation, 39(1-4):5--35, January 2000.

No context found.

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional performance modelling with the TIPPtool. Performance Evaluation, 39(1-4):5--35, January 2000.

.... to performance analysis has been illustrated by several examples, see e.g. 12,34,51,60] and important advances have been made in exploiting the structure of the compositionality for analysis purposes, for an overview see [64] Several algorithms have been implemented in tools, like the TIPPtool [48], PEPA Workbench [33] and TwoTowers (for EMPA) 13] For LOTOS preliminary proposals for stochastic extensions were presented by 10 Rico and von Bochmann [90] using semi Markov chains, and by Valderrutten et al. 102] who derived queueing networks from extended LOTOS specifications. A similar, ....

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional performance modelling with the TIPP-tool. In R. Puigjaner, N.N. Savio and B. Serra, eds, Computer Performance Evaluation, LNCS 1469, pages 51--63, 1998.

.... as its central underlying data structure [45,46] IM CAT is written in C and (similarly to PRISM) is built on top of the library CUDD [43] The main features of IM CAT are: 1) Reading of elementary MTSs from file in a simple format as generated by the stochastic process algebra tool TIPPtool [47] and generating their MTBDD representation. Actually, the Markovian transitions of an MTS are represented by an MTBDD and immediate transitions by a separate BDD, which later is turned into an MTBDD if non deterministic choice is resolved by probabilities as explained in Sec. 3.3. 2) Parallel ....

....2,097,152 7.33082e 06 485 16,383 5.36871e 08 1.8789e 09 677 Fig. 12. Statistics for the tandem queueing system (obtained with IM CAT) 7.3. 2 A tandem queueing network with blocking As a second example we consider a tandem queueing network with blocking taken from [51] and used again in [47]. It consists of a M Cox 2 1 queue sequentially composed with a M 1 queue, see Fig. 11. Each of the two queueing stations has a finite capacity of c jobs, c 0. Jobs arrive at the first queueing station according to a Poisson stream with rate . The service time of the first station has a ....

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, M. Siegle, Compositional performance modelling with the TIPPtool, Performance Evaluation 39 (1-4) (2000) 5--35.

....ones. After bisimilarity compression, the small model has 720 states and 3219 transitions, the intermediate model has 2640 states and 12295 transitions and the large model has 21648 states and 103471 transitions. All steady state properties given in the table were double checked with TIPPtool [22]. states (original) 3690 13530 110946 (compressed) 720 2640 21648 property verification runtimes (in seconds) Phi 1 0.012 0.037 0.268 Phi 2 0.008 0.049 0.864 Phi 3 0.008 0.039 0.319 Phi 4 0.003 0.005 0.036 Phi 5 0.642 2.371 18.750 Phi 6 0.001 0.002 0.014 Phi 7 0.558 2.122 18.814 ....

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis and M. Siegle. Compositional performance modelling with the TIPPtool. Performance Evaluation, 39(1-4): 5--35, 2000.

....ones. After bisimilarity compression, the small model has 720 states and 3219 transitions, the intermediate model has 2640 states and 12295 transitions and the large model has 21648 states and 103471 transitions. All steady state properties given in the table were double checked with TIPPtool [22]. 6 On translating aCSL to CSL The design of aCSL closely follows the work of De Nicola and Vaandrager on aCTL [33] For what concerns model checking, they propose a translation K from aCTL into CTL, and a transformation (also denoted K) from actionlabelled to state labelled transition ....

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis and M. Siegle. Compositional performance modelling with the TIPPtool. Performance Evaluation, 39(1-4): 5--35, 2000.

....Another important issue for industrial strength formal analysis is the availability of tool support. At the current state, prototypical tool support is available for both the stochastic modelling and the analysis phase: A couple of prototypes exist that allow a process algebraic modelling of CTMCs [19, 7, 5]. So far, performance models with up to 10 7 states have been modelled and analysed compositionally [20] A prototypical model checker for Markov chains, E T MC 2 , is also available [21] it was used to check the above CSL properties of the Hublle space telescope. More e#ort is nevertheless ....

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis and M. Siegle. Compositional performance modelling with the TIPPtool. Performance Evaluation 39(1-4):5--35, 2000.

No context found.

Hermanns, H., Herzog, U., Klehmet, U., Siegle, M., and Mertsiotakis, V. 1998a. "Compositional Performance Modelling with the TIPPtool". In: R. Puignajer, N. Savino, and Serra, B. (eds), Computer Performance Evaluation. LNCS 1469. Springer.

....The matrix R, together with the proposition labelling function L, constitutes the interface between the high level formalism at hand and the model checker E MC 2 . Currently, E MC 2 accepts CTMCs represented in the tra format as generated by the stochastic process algebra tool TIPPtool [22], but the tool is designed in such a way that it enables a filter plug in functionality to bridge to various other input formats. This is realized via Java s dynamic class loading capability. 3.1 The model checking algorithm Once the matrix R and the labelling L of a CTMC M have been ....

....and logical connectives is very fast. Checking steady state properties and unbounded until formulas is also a matter of only a few seconds, even for the 15360 state case. Measurements have shown that the performance of our tool s steady state solution algorithm is comparable to the one of TIPPtool [22] which is based on a sophisticated sparse ma (poll1 # poll2 ) P# #p (serve2 U serve1 ) busy1 # P#1 (#poll1 ) d # states time (in sec) time (in sec) time (in sec) 3 36 0.002 0.031 0.005 5 240 0.002 0.171 0.009 7 1344 0.005 1.220 0.011 10 15360 0.037 16.14 0.080 busy1 # P# #p (# ....

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis and M. Siegle. Compositional performance modelling with the TIPPtool. Perf. Ev., 39(1-4): 5--35, 2000.

....n denotes the number of states and m the cardinality of the transition relation. Weak Markovian bisimulation is determined using an adaptation of the algorithm of [5] Its time complexity is O(n 3 ) For the telephone system example we use the implementation of these algorithms in the TIPPtool [25]. 4 The plain old telephone system In this section we informally describe the functional behaviour of the plain old telephone system (POTS) and provide fragments of its formal specification in Basic LOTOS. Telephone systems have traditionally been a standard specification exercise in the context ....

....specification into pieces of moderate size. The constraint oriented character of the elapse operator has been of crucial importance here. For ag30 gregation purposes we applied ordinary strong and weak bisimulation (using C sar Ald ebaran [12,6] and stochastic variants thereof (using TIPPtool [25]) Although theoretically the worst case time and space complexity of these algorithms is the same [22] we experienced that the implementation of Ald ebaran outperforms that of TIPPtool to a significant extent. With respect to the state space generation, the same is true for C sar compared to ....

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional performance modelling with the TIPPtool. In R. Puignajer, N.N. Savio and B. Serra, eds, Computer Performance Evaluation, LNCS 1469, pages 51--63. Springer-Verlag, 1998.

No context found.

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional performance modelling with the TIPPtool. In R. Puigjaner, N. Savino, and B. Serra, editors, Proc. 10th International Conference on Modelling Techniques and Tools for Computer Performance Evaluation (TOOLS '98), volume 1469 of LNCS, pages 51-62. Springer, 1998.

No context found.

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional performance modelling with the TIPPtool. Perf. Eval., 39(1-4), 2000.

No context found.

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional performance modelling with the TIPPtool. Performance Evaluation, 39:5-35, 2000.

No context found.

H. Hermanns, U. Herzog, U. Klehmet, V. Mertsiotakis, and M. Siegle. Compositional performance modelling with the TIPPtool. Perf. Eval., 39(1-4), 2000.

First 50 documents

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