R. Grosu and K. Stlen. A denotational model for mobile point-to-point dataflow networks. Technical Report SFB3,B A, Technische Universitat Munchen, 1995.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... is given in [504, 890] Hardware specification is treated in [357, 356] A calculus for mobile data flow networks is given in [433, 150] A fully abstract, denotational model for mobile, timed, nondeterministic data flow networks whose components communicate in a point to point fashion is given in [439]. In this model the stream processing functions are time guarded and generic. Time guardedness assures unicity of fix points and genericity assures mobility. The relation between Focus agents and automata is studied in [438] Algebraic specifications and graph grammars. The algebraic approaches ....

R. Grosu and K. Stølen. A Denotational Model for Mobile Point-to-Point Dataflow Networks. Technical Report SFB 342/14/95 A, Technische Universitat Munchen, 1996. An extended abstract also occurs in Proc. AMAST'96. \Phi.

....thereby liveness) to be modeled in an elegant way. However, contrary to [Bro87] we use pulse driven functions and infinite timed streams. Thereby we get a simpler theory. The actual formulation of pulse drivenness has been taken from [Bro95] For more details about the semantic model we refer to [GS95]. Specification Formats: The formats for time dependent and time independent specifications are inspired by [BS94] The format for synchronous specifications is for example related to the approach in [Tuc92] The one step longer than semantics used by us is strongly inspired by [AL93] Col94] ....

R. Grosu and K. Stølen. A denotational model for mobile point-to-point dataflow networks. Technical Report SFB 342/14/95 A, Technische Universit at Munchen, 1995.

....Focus [BDD 93] Asynchronous system models provide the most abstract system model for systems with message exchange. They can easily be modeled using stream processing functions, for which a multitude of tractable specification techniques for untimed as well as for timed systems exist ( GS95] BDD 93] Moreover, for stream processing functions a powerful theory for compositional refinement has been developed. By using an asynchronous system model, in contrast to process algebraic approaches like the calculus [Mil91] or CCS [Mil89] we do not have to tackle synchronization ....

R. Grosu and K. Stølen. A Denotational Model for Mobile Point-toPoint Dataflow Networks. Technical Report TUM-I 9527, Technische Universitat Munchen, 1995.

....Step Semantics The underlying time hypothesis of our semantics is a global time measure. We assume that every Mini Statechart can make a step at least an idle loop at every single time tick. This assures time progress because every single transition takes place in exactly one time unit [GS95] This behaviour is guaranteed by reactive automata. Reactiveness requires the existence of at least one transition in every state of the automaton. The functionality of the strict step function is: stJ:K : S fin (M) fin (M) Theta S: The step semantics yields a finite set of output ....

R. Grosu and K. Stølen. A denotational model for mobile point-to-point dataflow networks. Technical Report SFB 342/14/95 A, Technische Universitat Munchen, Fakultat fur Informatik, 80290 Munchen, Germany, 1995.

....allow unbounded nondeterminism (and thereby liveness) to be modeled in an elegant way. However, contrary to [Bro87] we use pulse driven functions and infinite timed streams. Thereby we get a simpler theory. The actual formulation of pulse drivenness has been taken from [Bro95] We refer to [GS95] for more details on the semantic model. Specification Formats: The distinction between simple and general specifications can also be found in [SDW93] Bro94] However, in these papers, the techniques for expressing general specifications are more complicated. SDW93] uses a specification ....

R. Grosu and K. Stølen. A denotational model for mobile point-to-point dataflow networks. Technical Report SFB 342/14/95 A, Technische Universitat Munchen, 1995.

No context found.

R. Grosu and K. Stlen. A denotational model for mobile point-to-point dataflow networks. Technical Report SFB3,B A, Technische Universitat Munchen, 1995.

....by communicating channels and functions instead of ports. Bro95] and [Gro94] give also an equational characterization of dynamic reconfiguration. Mobility in the more general framework of nondeterministic systems and where reconfiguration is achieved by sending ports was studied thereafter in [GS95, GS96a, GSB97, GS96b] This article compares and unifies the models given in those papers. Moreover, it introduces for each communication paradigm a convenient specification formalism. This formalism was applied successfully to give a formal, high level specification of the kernel functionality ....

R. Grosu and K. Stlen. A Denotational Model for Mobile Point-to-Point Dataflow Networks. Technical Report SFB 342/14/95 A, Technische Universitat Munchen, 1995.

....earlier. 2 4. 2 Black Box View In many cases, for methodical reasons one is interested in a more abstract view of a component than the view given by a state machine or by a network (see the following section) In such cases, the black box view, also called the historybased approach, can be used [GS95b, GR95] In the history based approach we model components by sets of functions. Each function has the form f 2 (I [D ] O [D ] It maps named input histories to named output histories. The names of the input channels build the set I and the names of the output channels build the set ....

....j 1 = f( # j 1 (strongly pulse driven) We use the arrow for sets of strongly pulse driven functions and the arrow w for sets of weakly pulse driven functions. 2 Strongly and weakly pulse driven functions correspond to contractive and nonexpansive functions in the metric of streams (see [GS95b, GR95] As a consequence, by Banach s fixed point theorem, strong pulse drivenness guarantees unique fixed points of feedback loops. Theorem 3 Projection is a pulse driven function. Sum is a pulse driven function in each argument. Proof: # n and ( j O )# n only depend on # n and # n . 2 ....

[Article contains additional citation context not shown here]

R. Grosu and K. Stlen. A Denotational Model for Mobile Pointto -Point Dataflow Networks. Technical Report SFB 342/14/95 A, Technische Universitat Munchen, 1995.

....Focus [BDD 93] Asynchronous system models provide the most abstract system model for systems with message exchange. They can easily be modeled using stream processing functions, for which a multitude of tractable specification techniques for untimed as well as for timed systems exist ( GS95a] BDD 93] Moreover, for stream processing functions a powerful theory for compositional refinement has been developed. By using an asynchronous system model, in contrast to process algebraic approaches like the calculus [Mil91] or CCS [Mil89] we do not have to tackle synchronization ....

R. Grosu and K. Stlen. A Denotational Model for Mobile Pointto -Point Dataflow Networks. Technical Report TUM-I 9527, Technische Universitat Munchen, 1995.

....are generated by an inappropriate level of abstraction. Nondeterministic models are useful only if they contain enough information about the causality between messages. Abstracting too much from this information not only destroys compositionality but also reduces the expressive power. In [GS95] we proposed a timed model for the history based approach where components are described with sets of functions. This model is compositional and fully abstract with respect to external observations. Moreover, it is powerful enough 1 to describe not only nondeterministic fair merge components but ....

....but it also allows to handle unbounded nondeterminism. Note that the use of relations instead of sets of functions is problematic in connection with unbounded nondeterminism (see for example [Cos85, NP92] 4. 3 One to many Composition We define the composition slightly more general than in [GS95] because we relax the pulse driveness requirements and allow one to many communication. 15 F 1 F 2 O O 1 2 2 i o p I 1 I 1 I 1 O I O 2 1 O 2 1 O Figure 4.1: One to many composition Definition 14 (One to many composition) Given two components F 1 and F 2 with input channels in I 1 and ....

[Article contains additional citation context not shown here]

R. Grosu and K. Stlen. A Denotational Model for Mobile Point-to-Point Dataflow Networks. Technical Report SFB 342/14/95 A, Technische Universit at Munchen, 1995.

....The advantage of the first alternative is its simplicity with respect to formal reasoning and understanding. The advantage of the second alternative is that many things can be expressed more directly. In this paper we concentrate on the second alternative. The first alternative is investigated in [GS95b]. Point to point communication is guaranteed by imposing local constraints on the way 2 ports are communicated. Some readers may wonder why we at all find point to point communication interesting. After all, point to point communication is only a special case of many to many communication. The ....

R. Grosu and K. Stlen. A denotational model for mobile point-to-point dataflow networks. Technical Report SFB 342/14/95 A, Technische Universitat Munchen, 1995.

....with respect to formal reasoning and understanding. The advantage of the second alternative is that many things can be expressed more directly. However, the price to pay is a more complicated model. In this paper we concentrate on the first alternative. The second alternative is investigated in [GS96b] To keep the model simple components are not allowed to forward ports they receive on their input channels. Thus, we cannot change the communication partners of a component. However, we can build up new connections between components that are already connected. These new connections can be in ....

....A to B and a channel connecting B to C, but no channel connecting C to A. Now, suppose the component A creates a channel i; it keeps the receive port and sends the send port i to the component B, which again forwards i to C. We then obtain the network on the left hand side of Figure 2. In [GS96b] where we model the more general form of pointto point communication, we allow forwarding by constraining the functions to forget ports as soon as they are sent. A similar technique could have been used here. However, this would make the model more complicated. In fact, the whole advantage ....

R. Grosu and K. Stlen. A denotational model for mobile point-to-point dataflow networks with channel sharing. To appear as technical report, Technische Universitat Munchen, 1996.

....and since our strong guardedness constraint, guaranteeing the existence of a unique fix point, corresponds straightforwardly to contractivity. Because of the space limitations, we assume basic knowledge of metric spaces. For more details on metric spaces we refer to the full version of the paper [GS95] The full version also provides detailed proofs. The rest of the paper is organized as follows. Section 2 introduces basic notions like communication histories and stream processing functions. Section 3 formalizes the privacy invariants of mobile point to point systems. Section 4 introduces ....

R. Grosu and K. Stlen. A denotational model for mobile point-to-point dataflow networks. Technical Report SFB 342/14/95 A, Technische Universit at Munchen, 1995.

....in [ S 1 Omega S 2 ] In fact, the way [ is defined implies that for any specification S and functions , if 8 : 9 0 2 [ S ] 0 ( then 2 [ S ] Thus, the denotation of a specification is always closed in this sense. This closure property makes our model fully abstract [6] and simplifies the definitions of refinement. 7 In the sequel we distinguish between basic and composite specifications. The latter differ from the former in that they consist of several specifications composed by Omega . A time independent specification S j (i o) ti : R is said to be ....

....made in this paper. As explained in [4] behavioral refinement can be generalized to interface refinement by relating the concrete and abstract interface by a representation function in the style of [8] The three refinement principles proposed above can be generalized accordingly. We refer to [6] for a detailed investigation of the underlying semantic model. The principles of partial and total refinement were defined in [13] but in a less general setting. Conditional refinement is a straightforward generalization of behavioral refinement so straightforward that it seems unlikely that ....

R. Grosu and K. Stlen. A denotational model for mobile point-to-point dataflow networks. Technical Report SFB 342/14/95 A, Technische Universitat Munchen, 1995.

....channels and functions instead of ports. Bro95] and [Gro94] give also an equational characterization of dynamic reconfiguration. Mobility in the more general framework of nondeterministic systems and where reconfiguration is achieved by sending ports was studied thereafter in [GS95, GS96b, GSB97, GS96a] This article compares and unifies the models given in those papers. Moreover, it introduces for each communication paradigm a convenient specification formalism. This formalism was applied successfully to give a formal, high level specification of the kernel functionality of an operating ....

R. Grosu, K. Stlen, and M. Broy. A denotational model for mobile point-to-point dataflow networks with channel sharing. Technical Report TUM-I9717, Technische Universitat Munchen, 1997.

....by communicating channels and functions instead of ports. Bro95] and [Gro94] give also an equational characterization of dynamic reconfiguration. Mobility in the more general framework of nondeterministic systems and where reconfiguration is achieved by sending ports was studied thereafter in [GS95, GS96b, GSB97, GS96a] This article compares and unifies the models given in those papers. Moreover, it introduces for each communication paradigm a convenient specification formalism. This formalism was applied successfully to give a formal, high level specification of the kernel functionality ....

R. Grosu and K. Stlen. A denotational model for mobile point-to-point dataflow networks. Technical Report TUM-I9527, Technische Universitat Munchen, 1995.

....may be passed between programs. In higher order (HO) calculus, not only channel identifiers, but also whole processes may move along channels. This corresponds to passing clones in object oriented languages. As in CCS, communication is done synchronously using handshaking. DYNAMIC FOCUS [32, 33, 34, 35]: This approach extends FOCUS similar to the way calculus extends CCS by allowing channel names to be passed between processes. This leads to a new, denotational understanding of dynamics as a set of privacy invariants that have to be maintained by the interacting processes. As in FOCUS, ....

R. Grosu and K. Stlen, "A denotational model for mobile point-to-point dataflow networks with channel sharing," tech. rep., Technische Universitat Munchen, 1996.

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