Tải bản đầy đủ - 0 (trang)
4 Excursion: On the Voting Example

4 Excursion: On the Voting Example

Tải bản đầy đủ - 0trang

216



J. Desel and J. Esparza



a1



n

...

0



ak



y



y



n



reject1



n



...



...



...



rejectk



accept



a1



...

n

f



ak



Fig. 4. A corrected negotiation for unanimous vote.



transition is enabled if the acceptance place carries all k tokens, whereas for each

possible distribution of k tokens on the two places we define a separate rejecting

transition. As above, we might exclude the rejecting transition consuming k

tokens from the acceptance place. Formally, we define a place/transition Petri

net with arc weights. Apparently this net has linear size with respect to k, the

number of agents. The net is k-bounded, i.e., no reachable marking assigns more

than k tokens to a place.

Another possibility of modelling the same behaviour is to provide k places for

acceptance, one for each agent, and k places for rejection, as in our first Petri net.

Now we add, for each agent, a transition moving a token from the corresponding

acceptance place to the rejection place. This transition is labelled by the empty

word τ . No matter how the agents voted, we can reach the marking with all tokens

on the reject places by firing these τ -labelled transitions. Therefore, it suffices to

have one acceptance transition that removes tokens from all acceptance places

and one rejection transition that removes tokens from all rejection places. Firing

a τ -labelled transition does not contribute to the observed behaviour of the Petri

net. So this net is at least language equivalent to the negotiation of Fig. 3.

Summarizing, we have provided a systematic way to construct a 1-safe Petri

net corresponding to a negotiation, which can be exponentially larger than the

negotiation. For the voting example, this Petri net has exponentially many transitions. For this example we also provided a linear-sized Petri net with the same

behaviour, which is, however, not 1-safe but k-bounded. Another Petri net with

this behaviour is 1-safe, but has τ -labelled transitions. We actually do not know

if, for negotiations in general, there always exist polynomial-sized Petri nets

with the same behaviour which are 1-safe, which are bounded, which have no



Negotiations and Petri Nets



217



τ -labelled transitions etc., i.e. all these problems are open. For the voting example, we did not find a polynomial-sized equivalent 1-safe Petri net without τ labelled transitions.



4

4.1



Properties of the Net Associated with a Negotiation

S-components



An S-component of a Petri net is a subnet such that, for each place of the

subnet, all input- and output-transitions belong to the subnet as well, and such

that each transition of the subnet has exactly one input- and exactly one outputplace of the subnet [6]. It is immediate to see that the number of tokens in an

S-component never changes. A net is covered by S-components if each place

and each transition belongs to an S-component. Nets covered by S-components

carrying exactly one token are necessarily 1-safe. For example, every live and

1-safe free-choice net enjoys this nice property [6].

Petri nets associated with negotiations are not covered by S-components,

only because the end-transitions have no output places. However, if we add an

arc from each end-transition to each initially marked place, then the resulting

net is covered by S-components:

Proposition 4. The Petri net associated with a negotiation, with additional

arcs from each end-transition to each initially marked place, is covered by Scomponents.

Proof. We consider the Petri net with the additional arcs. For each agent a, the

subnet generated by all places [a, X] and all transitions labelled by (n, r) where

a ∈ Pn , is an S-component (being generated implies that the arcs of the subnet

are all arcs of the original net connecting nodes of the subnet). An arbitrary place

of the net belongs to one such subnet, because it corresponds to an agent. Each

transition has a label (n, r), and each atom n has a nonempty set of participants,

whence the transition belongs to the subnet of some agent.

4.2



Soundness



The following notion of sound negotiations was inspired by van der Aalst’s soundness of workflow nets [1]. It was first defined in [8].

Definition 9. A negotiation is sound if each outcome occurs in some initial

occurrence sequence and if, moreover, each finite occurrence sequence is a large

step or can be extended to a large step.

All the negotiations shown in the figures of this paper are sound. For an

example of an unsound negotiation, consider again the ping-pong negotiation

shown in Fig. 1 on the right hand side. Imagine that Daughter could choose to

start negotiating with Father or with Mother. This would formally be expressed

by replacing the arc from port D of n0 to port D of nFD by a hyperarc from port D



218



J. Desel and J. Esparza



of n0 to ports D of both nFD and nDM . If, in this modified distributed negotiation,

Daughter first negotiates successfully with Mother, a marking is reached where

both Daughter and Mother can only engage in the final atom nf , whereas Father

is still only able to participate in nFD . So the distributed negotiation has reached

a marking which is neither final nor enables any outcome. We call such a marking

a deadlock. Clearly, sound negotiations have no reachable deadlocks.

Since the Petri nets associated with negotiations are not workflow nets, we

cannot immediately compare the soundness notions of workflow nets and of negotiations. Instead we define additionally in/out-nets associated with negotiations,

which are obtained by a minor transformation from the originally constructed

Petri nets. These in/out-nets are a generalisation of workflow nets, as defined in

[1]. Soundness, as defined for workflow nets in [1], is generalized to in/out-nets

in the following definitions.

Definition 10. An in/out-net is a Petri net with two distinguished places pin

and pout such that pin has no input transition and pout has no output transition.

The initial marking of an in/out-net assigns one token to the place pin and

no token to all other places. In/out-nets also have a final marking, assigning one

token to pout and no token to all other places.

An in/out-net net is sound if it has no dead transitions (i.e., each transition

belongs to an initially enabled occurrence sequence) and, moreover, each initially

enabled occurrence sequence is a prefix of an occurrence sequence leading to the

final marking.

A workflow net is an in/out-net such that, for each place or transition x,

there are directed paths from pin to x and from x to pout .

Now we associate in/out-nets with negotiations.

Definition 11. The in/out-net associated with a negotiation is obtained from

the Petri net associated with the negotiation by the following transformations:

1. The Petri net associated with the negotiation has, for each participating agent,

an initially marked place. We delete all except one of these places and adjacent

arcs and rename the remaining initially marked place to pin .

2. We add an initially unmarked place pout and arcs from all transitions labelled

by outcomes of the final atom nf (which we called end before) to this place.

In/out-nets associated with negotiations are not necessarily workflow nets

because not every element is necessarily on a path from the initial place to the

final place. However, this condition holds if the negotiation is sound, as the

following proposition shows.

Proposition 5. The in/out-net associated with a sound negotiation is a workflow net.

Proof. By construction, the in/out-net has distinguished places pin and pout .

By definition of a distributed negotiation, the initial atom is not a possible

next atom for any atom and any agent, i.e., it does not belong to any X(n, a, r).



Negotiations and Petri Nets



219



Hence, by construction, the initially marked places of the Petri net associated

with the negotiation have no ingoing arcs. Since pin is one of these places, it has

no ingoing arc.

The new place pout has no outgoing arc.

Since, by soundness of the negotiation, every atom (and therefore every outcome) can be enabled, a token can be moved from the initial atom to any other

atom. Therefore, there is a directed path from the initial atom to any other

atom (more precisely, there is a path in the graph of the negotiation). By the

construction of the Petri net (and of the in/out-net), there are according paths

from the place pin to arbitrary places and transitions of the net.

Again by soundness of the negotiation, every occurrence sequence can be

extended to a large step, i.e., the final atom can eventually be enabled and the

final marking reached. So every token can be led to the final atom, and therefore

there are paths in the graph of the negotiation from every atom to the final

atom. By construction of the Petri net (and of the in/out-net), there are thus

paths from any element to an end-transition, where end is an outcome of nf ,

and – in the in/out-net – to the place pout .

Next we show that, for sound negotiations, the associated Petri net and

the associated in/out-net are behaviourally equivalent. To this end, we formally

introduce an equivalence relation on the set of Petri nets:

Definition 12. Two Petri nets N and N are in the relation R if

– N is obtained from N by the deletion of a place p and adjacent arcs and

– the reachability graphs of N and N are isomorphic.

The symmetrical, reflexive and transitive closure of R is called place equivalence.

Places p satisfying the condition of this definition are often called implicit.

Clearly, by construction place equivalence is an equivalence relation.

Lemma 3.

(a) Let N be a Petri net with two places p and p with identical sets of input

transitions, identical sets of output transitions and identical initial marking.

Then deletion of p together with adjacent arcs leads to a place-equivalent

net.

(b) Let N be a Petri net with a place p with no output transition. Assume that

there are no two distinct reachable markings m and m that disagree only

with respect to p, i.e., that satisfy

m(p) = m (p) and m(p ) = m (p ) for p = p .

Then deletion of p and adjacent arcs leads to a place-equivalent net.2

2



Without the second condition, i.e., assuming only that p has no output transitions,

the derived net is a bisimular net. It has in particular identical occurrence sequences

as the original one, but it can have a smaller reachability graph because distinct

reachable markings might differ only with respect to the place p.



220



J. Desel and J. Esparza



Proof.

(a) The nets N and N are obviously in the relation R as defined in Definition 12.

(b) Clearly, removing p does not change the behaviour in terms of occurrence

sequences because a place can only restrict the enabledness of its output

transitions, but p has no output transitions. The second assumption implies

moreover that, for each reachable marking m, the number m(p) follows

uniquely from all m(p ), p = p . So we have a bijective mapping from reachable markings of the Petri net N to reachable markings of the reduced net,

which is formally given by the projection of markings to the set of places

without p. It is easy to see that this bijection actually induces an isomorphism between the two reachability graphs.

Using this lemma we now show that, at least for sound negotiations, the

associated Petri net and the associated in/out-net have the same behaviour.

Proposition 6. Let N be a sound negotiation. The reachability graph of its associated Petri net is isomorphic to the reachability graph of its associated in/outnet.

Proof. As argued in the proof of Proposition 5, the initially marked places of

the Petri net associated with the negotiation have no ingoing arcs. Since the

initial atom of the negotiation has all agents as participants, the transitions corresponding to its outcomes consume the tokens from all initially marked places.

Therefore, all these places have the same (empty) set of input transitions and

the same set of output transitions. So Lemma 3(a) applies and proves that the

transformation leads to a net with identical reachability graph.

Next we show that adding the place pout also does not change behaviour. We

argue considering the in/out-net with the place pout and show that removing

this place leads to a net with isomorphic reachability graph. We aim at applying

Lemma 3(b), and thus have to show that no two distinct reachable markings of

the in/out-net differ only with respect to the marking of pout .

By construction of the Petri net and of the corresponding in/out-net associated to the negotiation, firing a transition labelled with an outcome of the final

atom removes all tokens from the net. This is because all agents participate in

the final atom. Conversely, these transitions are the only transitions which do

not produce tokens on some places. Therefore, there are tokens in the Petri net

before one of these transitions occurs and there are no tokens in the Petri net

afterwords. In particular, there can only be one occurrence of such a transition.

In the in/out-net, occurrences of transitions representing final outcomes add a

token to the place pout and no other transition changes the marking of this

place. Therefore, before the occurrence of a transition labelled by a final outcome there are marked places (one for each participant) and pout is unmarked.

After the occurrence of a transition labelled by a final outcome, pout is the only

marked place. So no two reachable markings differ only with respect to pout , and

Lemma 3(b) applies.



Negotiations and Petri Nets



221



Unfortunately, soundness of a negotiation does not necessarily imply soundness of the associated in/out-net (which is, by Proposition 5, a workflow net).

The reason is that soundness requires that every atom can occur but not that

every branch of a hyperarc is actually used. If, for example, there was an additional hyperarc in Fig. 1 from the port F in n0 to the ports F in nFD and nf instead

of the arc from n0 to nFD , then the resulting negotiation would still be sound;

actually, the behaviour does not change at all. The associated in/out-net, however, would have an additional transition end with new input place [F, {nFD , nf }]

(and other input places) which is never enabled. This net is therefore not sound.

4.3



Deterministic Negotiations



In [9], we concentrate on deterministic negotiations, which are negotiations without proper hyperarcs.

Definition 13. A negotiation is deterministic if, for each atom n, agent a ∈ Pn

and result r ∈ Rn , X(n, a, r) contains at most one atom (and no atom only if

n = nf ).

The term deterministic is justified because there is no choice for an agent

with respect to the next possible atoms.

Since both, the exponential blow-up and the problem of useless arcs (branches

of hyperarcs) stem from proper hyperarcs, we can expect that deterministic

negotiations allow for better results. Actually, the Petri net associated with a

deterministic negotiation is in fact much smaller, because all its places have the

form [a, X], where a is an agent and X is a singleton set of atoms. So the set of

places is linear in agents and in atoms.

Before discussing soundness of deterministic negotiations, we make a structural observation. For the definition of free-choice nets used here, see [6].

Proposition 7. The Petri net associated with a deterministic negotiation is a

free-choice net, i.e., every two places either share no output transitions, or they

share all their output transitions. The same holds for the in/out-net associated

with a deterministic negotiation.

Proof. Since, in a net associated with a deterministic negotiation, each place

has the form [a, X], where X is a singleton set {n}, all its output transitions are

labelled by (n, r), r being a possible result of n. By construction, every other

place [b, {n}] has exactly the same output transitions as [a, {n}] whereas all other

places have no common output transition with [a, {n}].

The transformations of Definition 11 do not destroy the free-choice

property.

Proposition 8. A deterministic negotiation is sound if and only if its associated

in/out-net is sound.

Proof. The translation from a negotiation to its associated Petri net can be

rephrased in a much simpler way if the negotiation is deterministic, as follows:



222



J. Desel and J. Esparza



– For each atom n and each a in Pn , we add a place [a, n].

– For each atom n and result r ∈ Rn , we add a transition (n, r) (no two transitions correspond to the same outcome (n, r), so transitions can be identified

with their previously used labels).

– Arcs connect all places [a, n] with all transitions (n, r).

– For each transition (n, r) with n = nf and each a ∈ Pn there is an arc from

(n, r) to [a, n ], where n is the unique atom in X(n, a, r).

– All places [a, n0 ] carry one token initially; all other places are initially

unmarked.

It is immediate that to see that the behaviour of the negotiation is precisely

mimicked by this Petri net. So the negotiation is sound if and only if the net has

no dead transitions and moreover can reach the final (empty) marking from any

reachable marking.

The result follows since the Petri net can, as above, be transformed into a

behaviourally equivalent in/out-net.

Combining Propositions 5 and 8 yields:

Corollary 4. If a negotiation is deterministic and sound then its associated

in/out-net is a sound workflow net.



5



From Petri Nets to Negotiations



In this section we study the converse direction: Given a labelled Petri net, is there

a negotiation such that the net is associated with the negotiation? Obviously,

for a positive answer the net has to enjoy all the properties derived before. In

particular, it must have disjoint S-components and initially marked input places.

However, in the general case it appears to be difficult to characterise nets that

have corresponding negotiations.

We will provide an answer for the case of sound deterministic negotiations

and sound free-choice workflow nets.

Proposition 9. Every sound free-choice workflow net is place equivalent to a

net which is associated with a sound deterministic negotiation.

Proof. A workflow net is sound if and only if the net with an additional feedback transition moving the token from pout back to pin is live and 1-safe [1]. Live

and 1-safe free-choice nets are covered by S-components [6]. Therefore a sound

free-choice workflow net is covered by S-components as well. However, these

S-components have not necessarily disjoint sets of places. Consequently, we

cannot easily find candidates for agents involved in the negotiation to be constructed.

Instead we proceed as follows: We choose a minimal set of S-components

that cover the net. Since each S-component of a live net has to carry a token, all

these S-components contain the place pin . Each S-component corresponds to an

agent of the net to be constructed. Each conflict cluster, i.e., each set of places



Negotiations and Petri Nets



223



sharing the same output transitions, corresponds a negotiation atom (remember

that the net is free-choice and therefore any two places either share all output

transitions or do not share any).

Each place p of the net is contained in at least one S-component of the cover.

Let Cp be the set of all S-components of the derived minimal cover containing

p. If Cp contains more than one S-component, we duplicate the place p, getting

a new place p with input and output transitions like p.

The new net still has a cover by S-components, where one of the

S-components containing p now contains p instead. Repetition of this procedure

eventually leads to a net where each place p belongs to exactly one S-component

Cp of the cover. Finally we delete the place pout . Both operations, duplication

of places and deletion of pout , lead to place-equivalent nets by Lemma 3.

The resulting net is associated with the following negotiation: The set of

agents is the set of S-components of the minimal cover. The atoms are the

conflict clusters of the net. The results of an atom are the transitions of the

corresponding conflict cluster. The X-function can be derived from the arcs of

the Petri net leading from transitions to places.



6



Conclusions



This contribution presented a translation from a distributed negotiation to a

behaviourally equivalent Petri net. The chosen notion of behavioural equivalence

is very strong, namely isomorphism of the reachability graphs.

In the worst case, the translation yields a Petri net exponentially larger than

the negotiation. We conjecture that this exponential blow-up is unavoidable, but

currently we do not have a proof. The problem of the succinctness of negotiations

with respect to weaker equivalence notions like bisimulation or language equivalence is also open. On the other hand, we have shown that for deterministic

negotiations the translation only causes a linear growth. Further, for deterministic negotiations soundness and non-soundness is respected by the transformation

to workflow-like Petri nets, whence in this case the reverse translation is possible

as well.

The translation to Petri nets is implicitly used in [8,9], and in a recent paper

on the analysis of Coloured Workflow Nets [11]. On the one hand, the fact that

deterministic negotiations are so closed to workflow free-choice nets guided our

efforts to obtain a reduction algorithm for the analysis of soundness and the

input/output relation of negotiations. On the other hand, in [11] we transferred

the reduction procedure back to Petri nets. The resulting reduction procedure

has been successfully applied to a collection of industrial workflows.

Since we do not currently have a large suite of negotiation models, while such

suites exist for workflow Petri nets, we have used negotiations mostly as a theoretical formalism to design new analysis techniques that can be later translated

to workflow nets. In future work we plan to analyze the connection between

negotiations and languages for the description of business processes. Negotiations could become an intermediate language between business processes and



224



J. Desel and J. Esparza



Petri nets, offering more compact descriptions and cleaner analysis procedures

and the possibility to apply the highly developed tool support for Petri net

analysis. This includes in particular model checking tools that can verify properties formulated in an appropriate Temporal Logic. Application of such tools to

negotiations requires not only prior transformation of the model but also of the

formula. So we are interested in appropriate languages for formalizing relevant

behavioural properties of negotiations.



References

1. van der Aalst, W.M.P.: The application of Petri nets to workflow management. J.

Circ. Syst. Comput. 08(01), 21–66 (1998)

2. Abdelzaher, T., Atkins, E.M., Shin, K.G.: QoS negotiation in real-time systems

and its application to automated flight control. IEEE Trans. Comput. 49(11),

1170–1183 (2000)

3. Basu, S., Bultan, T., Ouederni, M.: SIGPLAN notices. POPL 2012, 47(1) (2012).

issn = 0362-1340. ACM, New York

4. Chen, Y., Peng, Y., Finin, T., Labrou, Y., Chu, B., Yao, J., Sun, R., Willhelm, B.,

Cost, S.: A negotiation-based multi-agent system for supply chain management.

In: Proceedings of Agents 99 - Workshop on Agent Based Decision-Support for

Managing the Internet-Enabled Supply-Chain, pp. 15–20 (1999)

5. Davis, R., Smith, R.G.: Negotiation as a metaphor for distributed problem solving.

Artif. Intell. 20(1), 63–109 (1983)

6. Desel, J., Esparza, J.: Free Choice Petri Nets. Cambridge University Press, New

York (1995)

7. Emerson, E.A., Kahlon, V.: Rapid parameterized model checking of snoopy cache

coherence protocols. In: Garavel, H., Hatcliff, J. (eds.) TACAS 2003. LNCS, vol.

2619, pp. 144–159. Springer, Heidelberg (2003)

8. Esparza, J., Desel, J.: On negotiation as concurrency primitive. In: D’Argenio, P.R.,

Melgratti, H. (eds.) CONCUR 2013 – Concurrency Theory. LNCS, vol. 8052, pp.

440–454. Springer, Heidelberg (2013). Extended version in CoRR, abs/1307.2145,

http://arxiv.org/abs/1307.2145

9. Esparza, J., Desel, J.: On negotiation as concurrency primitive II: deterministic cyclic negotiations. In: Muscholl, A. (ed.) FOSSACS 2014 (ETAPS). LNCS,

vol. 8412, pp. 258–273. Springer, Heidelberg (2014). Extended version in CoRR,

abs/1403.4958, http://arxiv.org/abs/1403.4958

10. Esparza, J., Desel, J.: Negotiation programs. In: Devillers, R., Valmari, A. (eds.)

PETRI NETS 2015. LNCS, vol. 9115, pp. 157–178. Springer, Heidelberg (2015)

11. Esparza, J., Hoffmann, P.: Reduction rules for colored workflow nets. In: Stevens,

P., Wasowski, A. (eds.) FASE 2016. LNCS, vol. 9633, pp. 342–358. Springer, Heidelberg (2016). doi:10.1007/978-3-662-49665-7 20

12. Genest, B., Muscholl, A., Peled, D.A.: Message sequence charts. In: Desel, J.,

Reisig, W., Rozenberg, G. (eds.) Lectures on Concurrency and Petri Nets. LNCS,

vol. 3098, pp. 537–558. Springer, Heidelberg (2004)

13. Jennings, N.R., Faratin, P., Lomuscio, A.R., Parsons, S., Wooldridge, M.J., Sierra,

C.: Automated negotiation: prospects, methods and challenges. Group Decis.

Negot. 10(2), 199–215 (2001)

14. Peltz, C.: Web services orchestration and choreography. IEEE Comput. 36(10),

46–52 (2003). doi:10.1109/MC.2003.1236471



Negotiations and Petri Nets



225



15. Salaă

un, G., Bultan, T., Roohi, N.: Realizability of choreographies using process

algebra encodings. IEEE Trans. Serv. Comput. 5(3), 290–304 (2012)

16. Simon, C.: Negotiation Processes - The Semantic Process Language and Applications. Shaker, Aachen (2008)

17. Winsborough, W.H., Seamons, K.E., Jones, V.E.: Automated trust negotiation.

In: DARPA Information Survivability Conference and Exposition, DISCEX 2000

Proceedings, vol. 1, pp. 88–102. IEEE (2000)

18. Xu, H., Shatz, S.M.: An agent-based Petri net model with application to

seller/buyer design in electronic commerce. In: Fifth International Symposium

on Autonomous Decentralized Systems, ISADS 2001, Dallas, Texas, USA, 26–28

March 2001, pp. 11–18. IEEE Computer Society (2001)



A Formal Framework for Diagnostic Analysis

for Errors of Business Processes

Suman Roy1(B) and A.S.M. Sajeev2

1



Infosys Ltd., #44 Electronics City, Hosur Road, Bangalore 560 100, India

Suman Roy@infosys.com

2

Melbourne Institute of Technology, Sydney, NSW 2000, Australia

asmsajeev2@gmail.com



Abstract. Business process models expressed in languages such as

BPMN (Business Process Model and Notation), play a critical role in

implementing the workflows in modern enterprises. However, control

flow errors such as deadlocks and lack of synchronization, and syntactic

errors arising out of poor modeling practices often occur in industrial

process models. A major challenge is to provide the means and methods

to detect such errors and more importantly, to identify the location of

each error. In this work, we develop a formal framework of diagnosing

errors by locating their occurrence nodes in business process models at

the level of sub-processes and swim-lanes. We use graph-theoretic techniques and Petri net-based analyses to detect syntactic and control flowrelated errors respectively. While syntactic errors can be easily located on

the processes themselves, we project control-related errors on processes

using a mapping from Petri nets to processes. We use this framework to

analyze a sample of 174 industrial BPMN process models having 1262

sub-processes in which we identify more than 2000 errors. We are further able to discover how error frequencies change with error depth, how

they correlate with the size of the sub-processes and swim-lane interactions in the models, and how they can be predicted in terms of process

metrics like sub-process size, coefficient of connectivity, sequentiality and

structuredness.

Keywords: Verification · Formal methods · Processes · BPM Notation ·

Errors · Soundness · Petri nets · Workflow nets · Woflan · Diagnosis ·

Metrics



1



Introduction



Modern-day enterprises rely on streamlined business processes for implementing

the workflows in the operation. This is particularly important for internet-based

businesses where on-line processes such as accepting orders need to be seamlessly

integrated with physical processes like delivery of products. Correct implementation of process models can result in significant cost savings in industry. For

example, Hammer [Ham10] reports a computer manufacturer reducing time to

c Springer-Verlag Berlin Heidelberg 2016

M. Koutny et al. (Eds.): ToPNoC XI, LNCS 9930, pp. 226–261, 2016.

DOI: 10.1007/978-3-662-53401-4 11



Tài liệu bạn tìm kiếm đã sẵn sàng tải về

4 Excursion: On the Voting Example

Tải bản đầy đủ ngay(0 tr)

×