MODELLING, ANALYSIS AND IMPROVEMENT OF MOBILE

BUSINESS PROCESSES WITH THE MPL METHOD

Volker Gruhn and Andr

e K

ohler

University of Leipzig, Applied Telematics/e-Business Group, Klostergasse 3, 04109 Leipzig, Germany

Keywords:

Process modelling, Process analysis, Mobility.

Abstract:

This paper introduces the Mobile Process Landscaping (MPL) method for modelling, analysing and improving

mobile business processes. Current approaches for process modelling and analysis do not explicitly allow the

consideration of typical mobility issues, e.g. location-dependent activities, mobile networks as resources and

speciﬁcs of mobile information systems. Thus, our method focuses on the modelling and analysis of these

characteristics, and is furthermore based on the process landscaping approach, supporting the easy creation

of hierarchical models of distributed processes. The method comes with a specialized modelling notation and

guidelines for the creation of process landscapes, context models, and business object models. Furthermore,

it provides a catalogue of formally deﬁned evaluation objectives, targeting at typical mobility issues. Each

evaluation objective can automatically be tested on the created process landscape. Furthermore, the method

includes a best practices catalogue with patterns for process and application improvements for typical mobility

situations. A validation of the method is presented showing results from the method’s use in a real-world

project.

1 INTRODUCTION

The subject of mobile business processes is to per-

form on-site services, either for external partners of

a company or due to internal requirements. Current

approaches for modelling, analysing and improving

mobile business processes and the supporting mo-

bile applications do not satisfy current industry re-

quirements. This paper introduces the Mobile Pro-

cess Landscaping (MPL) method in order to close this

methodological gap. It leads its user from mobility-

speciﬁc evaluation objectives in the beginning to tan-

gible process and application improvements in the

end. The method includes a detailed procedure de-

scription, a modelling notation, modelling guidelines,

an evaluation objectives catalouge, the deﬁnition of

analysis algorithms, support for result interpretation,

and a best practices catalouge for mobile process and

applications improvement.

The paper is organized as follows. The method

along with a case study from the insurance industry

is presented in section 2. In section 3, an overview

about the related work is given. Section 4 contains a

summary and draws a conlusion.

2 THE MPL METHOD

Within this chapter, the MPL method is described in

more detail. For this purpose it is shown how the

method was applied in a real-world case, taken from

the insurance industry. Subject of this case study is

the assessment of vehicle damages after car accidents

by an assessor of an insurance company.

In our case study, the company was not satisﬁed

with the overall performance of the process. Within

this situation, the MPL method was applied. In the

following sections the four steps of the method are

explained by example of this case:

1. Creation of a mobile business process model.

2. Selection of evaluation objectives.

3. Model analysis regarding the choosen evaluation

objectives.

4. Interpretation of analysis results and deduction of

appropriate activities.

Gruhn V. and Köhler A. (2009).

MODELLING, ANALYSIS AND IMPROVEMENT OF MOBILE BUSINESS PROCESSES WITH THE MPL METHOD.

In Proceedings of the 11th International Conference on Enterprise Information Systems - Information Systems Analysis and Speciﬁcation, pages 89-94

DOI: 10.5220/0001964400890094

 SciTePress

class Role model: assessor

«role»

assessor

«device type»

notebook

«network type»

DSL

«application»

ClaimsM anager

«application»

TaskManager

«application»

ExternalWebsite

Figure 1: Role model: assessor.

class Location model: task location

«location»

task location

GPRS

T-Mobile GPRS

::GP RS

- bandwidth_up_max_kb: int = 20

- bandwidth_down_m ax_kb: int = 56

- distribution_param eter_q: int = 1,2

- distribution_param eter_p: int = 1,6

UMTS

T-M obile UM TS

::UMTS

- bandwidth_up_max_kb: int = 128

- bandwidth_down_max_kb: int = 256

- distribution_param eter_q: int = 1,2

- distribution_param eter_p: int = 1,6

Figure 2: Location model: task location.

2.1 Method Step 1: Creation of a

Mobile Business Process Model

The model created during the use of the MPL method

is called the mobile business process model. It con-

sists of a process context model, a business object

model and a process landscape model.

2.1.1 Process Context Model

A mobile business process is executed within a con-

text, which is of particular relevance for the process

evaluation. Such context information cannot be mod-

elled as a process structure. Thus, a separate model is

created which contains:

• roles that a person can act as,

• locations at which a process step is executed,

• applications that a role can use within in a process

step,

• devices necessary for running the applications,

• services provided by an application, and

• networks available at some location and required

by applications.

The MPL method provides a metamodel for the cre-

ation of the process context model, which is consti-

tuted as UML class diagrams (Figure 1 and 2 show

examples).

2.1.2 Business Object Model

The construction of business object models is a well-

known approach and will not be described here in de-

tail (see e.g. (Eeles and Sims, 1998)). The model

should contain all business objects exchanged within

the process as well as their relationships to each other.

It is realized as one or more UML class diagrams.

2.1.3 Process Landscape Model

During earlier work, we introduced the process land-

scaping approach (Gruhn and Wellen, 2001), (Gruhn

and Wellen, 2000). It pays particular attention to the

modelling of distributed processes with emphasis on

the data and documents moving around between dif-

ferent locations. The strength of this approach is that

it consists of hierarchically structured process models

and interfaces between them, which can be described

at different levels of detail. It turned out that this ap-

proach is very useful for modelling mobile business

processes. Thus, it is used in the presented method.

A mobile process landscape is modeled as an

overview of the mobile business processes of an or-

ganization. A process landscape is realized as one

process model, containing process steps that can be

reﬁned by other process models or application inter-

action models on demand. A process model describes

what the mobile worker is doing (business view). An

application interaction model describes, how the mo-

bile worker is using mobile technologies within a pro-

cess step (application view). As modelling notation

for process landscapes coloured petri nets (CPN) are

used (Jensen et al., 2007). They allow the graphical

modelling of control and object ﬂows, have a strong

formal basis and are thus ideal for static and dynamic

analysis purposes. Furthermore, a couple of standard

analysis techniques are available.

A cut-out of the process landscape model for the

insurance industry case is given in Figure 3, showing

the top-level of the process landscape.

Each activity in this model is reﬁned by another pro-

cess model, describing the reﬁned activity in more de-

tail. Within a process model the attributes role and

location must be speciﬁed for each process step, in-

dicating which role is executing the process step at

which possible locations. For each application inter-

ICEIS 2009 - International Conference on Enterprise Information Systems

[]

postprocessing

phase

PL5

(*activityType: ProcessStep*)

postdispatching

phase

PL4

(*activityType: ProcessStep*)

execution

phase

PL3

(*activityType: ProcessStep*)

TASK

dispatched

tasks

TASK

executed

tasks

1`[]

TASKLIST

task list ready for

execution

TASKLIST

task ready for

pre-dispatching

TASK

PL3

PL4

PL5

predispatching

phase

PL2PL2

preprocessing

phase

PL1PL1

finished

tasks

customer claims

notification

160`(99, 77, "new")

Figure 3: Root model for case study.

action, the attribute required connection class must be

speciﬁed, indicating which type of network connec-

tion is required. The attribute values are necessary for

the later model evaluation and must match the before

described context model.

Regarding the claims assessor case, it turned out

that the creation of consistent context, business object

and process landscape models required signiﬁciant

modelling effort. The reason was the unexpectedly

different perception of the process context, business

terms, and process structures in the project group.

The resulting discussion led to a signiﬁcantly better,

more detailed and commonly agreed understanding of

the process and its contexts. This knowledge gain, re-

sulting from the MPL method use, was a nice addi-

tional beneﬁt.

2.2 Method Step 2: Selection of

Evaluation Objectives

The second step in the MPL method is the selection

of evaluation objectives. Current process modelling

methods and tools offer a large number of general

evaluation objectives, e.g. identiﬁcation of deadlocks,

measuring time and costs and many more (Giaglis,

2001), (Hommes and Van Reijswoud, 2000).

However, to achieve as meaningful results as possible,

the speciﬁc process characteristics need to be consid-

ered during process evaluation. Compared to non-

mobile business processes, mobile ones are usually

more difﬁcult to manage in certain aspects. They are

expensive due to mobility costs and complicated to

plan because of place, time and resource restrictions.

Furthermore, mobile processes often do not possess

the necessary ﬂexibility for an appropriate reaction to

suddenly changed situations on-site. Thus, the MPL

method provides a catalogue of evaluation objectives,

purpose-built for the analysis of mobile business pro-

cesses (see Table 1).

Table 1: Evaluation objectives catalogue.

EO.1 Process Distribution Objectives

EO.1.1 Distribution by execution location

EO.1.2 Distribution by executing role

EO.2 Process Performance Objectives

EO.2.1 Mobile idle periods of information ob-

jects

EO.2.2 Required time to process completion

EO.3 IT-support Objectives

EO.3.1 Availability of applications and data

EO.3.2 Timeliness of applications and data

EO.3.3 Process robustness in ofﬂine situations

Table 2: Excerpt from evaluation objectives catalogue

(by example of EO.1.1).

Objective identiﬁer. EO.1.1 Distribution by exe-

cution location

Objective motivation. Different process parts

are executed at different locations. Each loca-

tion change causes efforts and requires time. The

more location changes occur within a process, the

higher is its potential for optimzation.

Objective goal. To evaluate the fragmentation of

the process model regarding location changes.

Model preparation. A location transition graph

is required.

Model evaluation. To identify sequences in the

location transition graph, containing the same mo-

bile location node, separated by one or more sta-

tionary nodes, at least two times.

Evaluation result. List with sequences fulﬁlling

the above given deﬁnition.

Usage of evaluation result. It should be tried to

avoid the identiﬁed sequences by manual process

optimization.

For each evaluation objective (EO), the catalogue

contains an in-depth description of the evaluation

MODELLING, ANALYSIS AND IMPROVEMENT OF MOBILE BUSINESS PROCESSES WITH THE MPL METHOD

goal, a description of how to perform the model anal-

ysis, and a description of the expected results along

some interpretation guidelines. A detailed presenta-

tion of the catalogue itself is beyond the scope of this

paper. Nonetheless, Table 2 gives an impression of

how an objective is deﬁned by example of EO.1.1.

2.3 Method Step 3: Model Analysis

Regarding the Choosen Evaluation

Objectives

Once the business process model is created and the

desired evaluation objectives are choosen, the model

evaluation can start. As a real-world model consists

easily of a couple of hundred activities, distributed

over hierachical submodels, the evaluation can hardly

be executed manually. Thus, appropriate tool support

comes with the MPL method, allowing to

• check the CPN model against the MPL modelling

guidelines,

• check integrity between context, business object,

and process landscape models, and

• check the CPN model against evaluation objec-

tives.

However, the scope of this paper is not to illustrate the

tool support, but to present how the process model

can be analysed. This will be done by example of

the claims assessor case. For the sake of brevity, the

analysis results are described only for the evaluation

objectives EO.1.1. As described in the deﬁnition of

these analysis (see Table 2), the creation of a loca-

tion transition graph is required. A location transition

graph is a reduction of the original process model. A

set of connected activities from the process model ex-

ecuted at the same location is reduced to a single node

in the location transition graph. Each arc between

these nodes stands for a location change.

Figure 4 shows the location transition graph for

the claims assessor case. A node in the location tran-

sition graph can represent a stationary location (S L) or

a mobile location (ML). Nodes with the same label in

the location transition graph represent different parts

of the process model, their included activities are to

be executed at the same location.

The ﬁrst node SL

stands for a group of connected

activities to be executed in the backofﬁce of the com-

pany (preprocessing phase). The following node SL

stands for a group of activities to be executed in the

dispatcher’s ofﬁce (predispatching phase), node S L

contains the activities to be executed in the home of-

ﬁce of the mobile worker (part of execution phase).

These nodes represent activities from different sta-

tionary locations, whereas node ML

represents ac-

tivities from the task location, which is a mobile lo-

cation (execution phase). This model part is followed

by a another group of activities to be executed in the

home ofﬁce of the mobile worker. Subsequently, a

group of activities to be executed at the dispatchers of-

ﬁce (postdispatching phase) and the backofﬁce (post-

processing phase) follow. Arcs between these nodes

represent possible locations transitions in the process

model.

The subject of the analysis EO.1.1 is to evaluate

how strong the process model is fragmented regarding

location changes (see Table 2). The evaluation goal is

to identify sequences in the location transition graph,

which contain the same mobile location node, sepa-

rated by one or more stationary nodes, at least two

times. The results of analysis EO.1.1 are illustrated

in Figure 4. Two sequences were found. The ﬁrst

sequence shows the case that after task execution the

task is getting reassigned by the dispatcher. This can

be necessary, if the task state or the produced result

does not match the dispatchers requirements. This re-

sults in the fact that the mobile task passes the model

area ML

a second time. The second sequence has a

similar reason. The customer agent can also reassign

the task to the mobile worker, due to the same reason

and with same consequences.

This analysis identiﬁed two weak points in the

process model, which are typical mobility problems.

These recurrent visits are necessary, if the mobile

worker can not accomplish his task due to missing

ressources or suddenly changed requirements at the

point of service. Recurrent visits cause usually high

costs and a long process duration and should thus be

avoided if possible.

2.4 Method Step 4: Interpretation of

Analysis Results and Deduction of

Appropriate Activities

The MPL method does not leave the user alone with

the evaluation results, but provides a catalogue of best

practices for changing process structures and the sup-

porting applications in order to improve the mobile

business processes regarding the incurring process

costs, the required time for process completion and

the quality of the process outcome. The best practices

are extracted from related work and are described in

the well-known pattern approach (see, e.g., (Rising,

1998)).

To present this catalogue is beyond the scope of

this paper. Nonetheless, the induced activities regard-

ing the above described evaluation results will be il-

lustrated. Regarding the situation of the recurrent vis-

ICEIS 2009 - International Conference on Enterprise Information Systems

backoffice

dispatcher office

home office

task location

# Sequence

1 ML

→ SL

→ ML

2 ML

→ SL

→

Figure 4: Location transition graph and evaluation results for EO.1.1.

its (EO.1.1), an analysis of the activating transitions

lead to the result, that the mobile worker may not be

able to ﬁnd the damaged car’s location due to missing

informations within the task description, especially if

the car’s location is different from the address of the

insurance holder. Thus, the application used for the

generation of the tasks was modiﬁed to include the

customer’s home address as well as the car’s loca-

tion address. The recording of the latter was inserted

as an activity into the initial service agent’s process.

Through this addition, the incidence rate of recurrent

visits could be reduced by nearly half.

This illustration is just a small section of the whole

analysis result and the induced improvement actions.

The results caused an extensive and detailed discus-

sion about possible process and application improve-

ments in the company. In conclusion, a bundle of ac-

tions was created to apply a large number of the dis-

cussed changes. Finally, after completion of these ac-

tions, the company could observe signiﬁcant improve-

ments in process costs and time. Thus, it could be

shown that the use of the MPL method was beneﬁcal

for the improvement of mobile business processes in

this case.

3 RELATED WORK

Barnes gives a taxonomy as well as examples for mo-

bile distributed work (Barnes, 2004). The taxonomy

framework allows to classify mobile business cases.

It consists of the dimensions mobility (transient, mo-

bile, remote), process (automation, decision support,

transformation) and value proposition (mobile chan-

nel access, mobile service value, mobile service cre-

ation). This framework is a good starting point if a

company begins with a ﬁrst evaluation of their mo-

bile business processes. Innes et al. present an appli-

cation of this framework for trade services organisa-

tions (Innes et al., 2005).

Yuan and Zheng also provide a framework for

analysing mobile work support (Yuan and Zheng,

2005). The framework consists of the dimensions

worker, task, context, and technology. With this

framework characteristics of stationary and mobile

work are compared in order to better understand mo-

bile work situations. An extension of this frame-

work is presented in (Yuan and Zheng, 2006), focus-

ing on the ﬁt between mobile task and mobile tech-

nology types. Perry et al. (Perry et al., 2001) also

present results regarding the basic concept of mo-

bility. They present a study where different aspects

of mobile work in general are examined. The four

key factors identiﬁed for mobile work are the role of

planning, working in dead time, accessing remote re-

sources and monitoring distant activities.

Giaglis proposes an evaluation framework for

business process modelling and information systems

modelling techniques (Giaglis, 2001). It consists of

three evaluation variables: the breadth (typical mod-

elling goals and objectives), the depth (functional, be-

havioral, organizational, and informational perspec-

tive), and the ﬁt (typical project types the technique

could support). When applying this framework to

the MPL method, the result is that the method has

a limited operational area (i.e. process development

and improvement). Furthermore, it turned out that

the MPL method supports the whole depth dimension

(support of functional, behavioral, organizational, and

informational perspective). Combining these results,

the ﬁt dimension gives projects types where the MPL

method could be used. Despite others they are: sys-

tem analysis and design, business process reengineer-

ing, task redesign, and workﬂow design.

A comprehensive study of methodologies, tech-

niques, and tools for business process change is pre-

sented by Kettinger et al. (Kettinger and Teng, 1997).

The authors examined nearly all available methods

and tools in an extensive research project. The main

result is a classiﬁcation scheme, suitable for discov-

ering the right method or tool for a certain purpose.

Within this scheme, the MPL method belongs to the

stages 3: Diagnose (document existing process, an-

alyze existing process), and 4: Redesign (deﬁne and

analyze new process concepts, prototype and detailed

design of new process, analyze and design informa-

tion systems).

MODELLING, ANALYSIS AND IMPROVEMENT OF MOBILE BUSINESS PROCESSES WITH THE MPL METHOD

4 SUMMARY AND

CONCLUSIONS

This paper presented the MPL method, which aims

to model, analyse and improve mobile business pro-

cesses. We have shown that current approaches do

not consider mobility adequately. But, since many

industries have to deal with such processes, a spe-

cialised method is required. Therefore, we presented

the MPL approach that provides a catalogue of pre-

deﬁned evaluation objectives and support for the cre-

ation of appropriate models. The evaluation of the

resulting model can be automated, leading to substan-

tial results that show possible process and application

improvements. The method was applied to a process

improvement project from the insurance industry. It

was shown that the intended beneﬁts could be real-

ized. Nonetheless, the method has also a couple of

limitations. Firstly, the modelling requires a person

having strong skills in UML and CPN. Secondly, the

quality of the produced results depends heavily on the

model quality. An experienced modeler with the eval-

uation objectives catalogue in mind will most likely

produce good models, while an unexperienced ﬁrst-

time modeler may not. Thirdly, the method produces

- as all process modelling approaches - a relatively in-

ﬂexible model due to the nature of sequences, decsion

nodes, and so on. But especially mobile business pro-

cesses usually allow a great ﬂexibility at the point of

service, i.e., being able to do anything anytime and

anywhere. Further research is planned to extend this

method. Currently, the tool support is still in the pro-

totype stage. The development of a more convenient

tool would be useful. Also, the evaluation objectives

catalogue has potential for extension.

ACKNOWLEDGEMENTS

The Applied Telematics/e-Business Group is en-

dowed by Deutsche Telekom AG. We would like to

thank Matthias Book and Ralf Laue for their valuable

comments on this paper.

REFERENCES

Barnes, S. J. (2004). Wireless support for mobile distributed

work: A taxonomy and examples. In Proceedings

of the Proceedings of the 37th Annual Hawaii In-

ternational Conference on System Sciences, pages

30078.1+. IEEE Computer Society.

Eeles, P. and Sims, O. (1998). Building business objects.

John Wiley & Sons, Inc., New York, NY, USA.

Giaglis, G. M. (2001). A taxonomy of business process

modeling and information systems modeling tech-

niques. International Journal of Flexible Manufac-

turing Systems, 13(2):209–228.

Gruhn, V. and Wellen, U. (2000). Structuring complex

software processes by process landscaping. In Soft-

ware Process Technology, Lecture Notes In Computer

Science, Vol. 1780, pages 138–149, Heidelberg, Ger-

many. Springer.

Gruhn, V. and Wellen, U. (2001). Process landscaping:

Modelling distributed processes and proving proper-

ties of distributed process models. In Unifying Petri

Nets, Lecture Notes In Computer Science, Vol. 2128,

pages 103–125. Springer, London, UK.

Hommes, B. J. and Van Reijswoud, V. (2000). Assessing

the quality of business process modeling techniques.

In Proceedings of the 33rd Hawaii International Con-

ference on System Sciences, pages 1007+. IEEE Com-

puter Society.

Innes, D., Barnes, S. J., and Scornavacca, E. (2005). The

impact of wireless ﬁeld force automation on new

zealand trade services organizations. In Proceedings

of the International Conference on Mobile Business,

pages 49–55. IEEE Computer Society.

Jensen, K., Kristensen, L., and Wells, L. (2007). Coloured

petri nets and cpn tools for modelling and validation

of concurrent systems. International Journal on Soft-

ware Tools for Technology Transfer (STTT), 9(3):213–

254.

Kettinger, W. J. and Teng, J. T. C. (1997). Business pro-

cess change: a study of methodologies, techniques,

and tools. MIS Quarterly, 21(1):55–80.

Perry, M., O’Hara, K., Sellen, A., Brown, B., and Harper, R.

(2001). Dealing with mobility: understanding access

anytime, anywhere. ACM Transactions on Computer-

Human Interaction, 8(4):323–347.

Rising, L., editor (1998). The Patterns Handbook: Tech-

niques, Strategies, and Applications. SIGS Reference

Library. Cambridge University Press.

Yuan, Y. and Zheng, W. (2005). From stationary work

support to mobile work support: A theoretical frame-

work. In Proceedings of the International Conference

on Mobile Business, pages 315–321. IEEE Computer

Society.

Yuan, Y. and Zheng, W. (2006). The ﬁt between mobile

task and mobile work support: A theoretical frame-

work. In Proceedings of the International Conference

on Mobile Business, pages 11+. IEEE Computer So-

ciety.

ICEIS 2009 - International Conference on Enterprise Information Systems