Graphical user interface

The graphical user interface manages the complete user interaction. It supports:

• specification of QPN models,
• specification of analysis tasks, selection of analysis algorithms and required performance measures,
• presentation of results concerning classification of QPNs and results of qualitative and quantitative analysis
• aggregation of detailed information on probably undesired net properties, e.g. brief description of a firing sequence which leads to a deadlock.

The model description process consists of several main steps.

1.

The net's graphical representation is clearly dominated by the Coloured Petri net part. This part is specified by creating and positioning places and transitions and establishing their connections by directed arcs. Figure 3 shows the net corresponding to Example 1.

2.

Attributes of places and transitions have to be set. This includes entering the different colours in a list, choosing the type of transitions and places (timed or immediate, resp. timed or ordinary) and fixing the number of tokens for the initial marking. Especially timed places require some additional information:

1. for each colour
   1. its rank according to bulk arrivals
   2. its service time distribution specified by its mean and coefficient of variance. This service time distribution is approximated by a Cox distribution as described in [8].
2. scheduling strategy
   Presently available scheduling strategies are FCFS, LCFS-Pr, PS and Infinite Server (IS).
3. number of servers
4. performance figures
   to be determined in quantitative analysis (cf. Sec. 3.2.4).

3.

The incidence functions of the Coloured Petri net (cf. [10]) have to be specified. This is possible in a graphical submodel at each transition. Such a submodel describes the locally unfolded net regarding a single transition and its input and output places. A special feature of QPN-Tool fills the submodel automatically with the transition's colours and all input and output places including all of their colours. Thus only arcs and their weights have to be specified manually.

Figure 6 shows the submodel of transition Fork in Fig. 3. A rhomb represents a coloured place. Colours of a place are represented by circles which are connected to its corresponding place by lines. Bars display colours of the transition whose submodel is regarded. Obviously transition colour `fork_light' takes one `light' token from place `Station_1' and puts on places `Wait' and `Station_2' one token on each of the corresponding colour `light'. Transition colour `fork_heavy' does the same for `heavy' tokens.

Locally unfolding a net has certain advantages:

• It allows a detailed view on single active components with their corresponding environments.
• The Petri-net-type formalism in locally unfoldings and the whole net is homogeneous.
• Automatic generation of graphical objects in a local unfolding supports a convenient specification.
• Furthermore generating available colours of input/output places as well as transition colours automatically avoids inconsistent specifications, because the set of colours which can be used and should be used are presented completely. Default positions reflect the common reading direction: input places and their colours to the left, transition colours in the middle, output places and their colours to the right.
• If several transitions have identical submodels, it is sufficient to specify just one. Sharing a submodel is possible as well as copying it.