A Novel Approach and a Language for Facilitating Collaborative

Production Processes in Virtual Organizations Based on DLT Networks

Nikola Todorovi

1 a

, Marko Vje

stica

1 b

, Nenad Todorovi

1 c

, Vladimir Dimitrieski

1 d

and Ivan Lukovi

2 e

University of Novi Sad, Faculty of Technical Sciences, Trg Dositeja Obradovi

ca 6, 21000 Novi Sad, Serbia

University of Belgrade, Faculty of Organizational Sciences, Jove Ili

ca 154, 11000 Belgrade, Serbia

Keywords:

Virtual Organizations, Process Modeling, Distributed Ledger Technology, Blockchain, Smart Contracts,

Internet of Things.

Abstract:

Due to strong competition and rapidly shifting market conditions, it is becoming harder for Small and Medium-

sized Enterprises (SMEs) to achieve business success. To deal with rising challenges, SMEs form Virtual Orga-

nizations (VOs) and seize business opportunities jointly. In this paper, we present an outline of a novel method-

ological approach that promotes trustworthy collaborative production execution within a non-hierarchical VO.

Furthermore, we propose using Distributed Ledger Technology (DLT) platforms and smart contracts to fa-

cilitate VO integration. The approach is based on the MultiProLan Domain-Speciﬁc Modeling Language

(DSML) extended with concepts required to allow process designers to (i) model collaborative production

processes while preserving the conﬁdentiality of private enterprise data and (ii) conﬁgure what data should be

shared between participants during the collaborative production execution. Designed process models are used

to automatically generate smart contracts by following the Model-Driven (MD) principles. Finally, generated

smart contracts are stored in a DLT network and used to distribute production data between VO participants and

monitor production execution in near real-time. The application of our methodological approach is demon-

strated by showcasing the use of the Collaborative Extension of MultiProLan (CE-MultiProLan) modeling

language and its concepts for modeling collaborative production processes.

1 INTRODUCTION

A Virtual Organization (VO) represents a temporary

alliance for integrating competencies and resources

from several independent collaborative companies to

satisfy customer’s requirements, or seize business op-

portunities by jointly developing complex products

(Priego-Roche et al., 2009). VOs are mainly formed

by SMEs because their unique capabilities are no

longer sufﬁcient for them to individually compete

with large companies and countries with lower la-

bor costs (Shamsuzzoha et al., 2013). In this con-

text, Non-Hierarchical Networks (NHNs) represent

VOs formed by companies of similar product port-

folios and sizes where SMEs enjoy equal rights and

https://orcid.org/0000-0001-8850-8439

https://orcid.org/0000-0003-2368-5818

https://orcid.org/0000-0002-9809-5719

https://orcid.org/0000-0003-3234-6543

https://orcid.org/0000-0003-1319-488X

controlling power (Shamsuzzoha et al., 2016).

There have been various efforts to assist SMEs

in forming and operating NHNs for the collabora-

tive development and delivery of customized prod-

ucts. In (Carneiro et al., 2010), authors present re-

sults of a business requirements analysis they con-

ducted in cooperation with industry partners. The

goal of the analysis was to determine what major re-

quirements should be addressed to promote collabo-

ration in NHNs. Among the most important require-

ments, they listed: (i) selection of partners for a spe-

ciﬁc business opportunity, (ii) standardization and im-

provement of communication within a VO, and (iii)

updating production statuses.

The existing solutions address these requirements

by integrating participants’ IT systems to enable shar-

ing data about events of interest during production

planning and execution, gathered by relying on the

Internet of Things (IoT). Here, IoT refers to the net-

worked interconnection of devices that can be used

in production, e.g., Radio Frequency IDentiﬁcation

Todorovi

c, N., Vještica, M., Todorovi

c, N., Dimitrieski, V. and Lukovi

c, I.

A Novel Approach and a Language for Facilitating Collaborative Production Processes in Virtual Organizations Based on DLT Networks.

DOI: 10.5220/0010720900003062

In Proceedings of the 2nd International Conference on Innovative Intelligent Industrial Production and Logistics (IN4PL 2021), pages 197-208

ISBN: 978-989-758-535-7

197

(RFID) tags, sensors, actuators, and similar. How-

ever, on the one hand side, these solutions do not pro-

mote transparent cooperation, i.e., they do not put an

emphasis on providing a deeper insight into how a VO

participant executes a speciﬁc production operation

and if the production is executed in accordance with

the contracted speciﬁcation. Consequentially, within

these solutions, records of events that occur during

production are still primarily stored and maintained

within isolated IT systems of participants that execute

a speciﬁc production operation. A solution that facil-

itates VOs must promote transparent cooperation be-

tween participants of a VO because participants may

be subjected to a liability action for faults in their per-

formance as part of a VO. A lack of transparency

could make it almost impossible to determine who

performed a particular action, resulting in uncertainty

regarding legal liability. Correct attribution of liabil-

ity should be facilitated by providing precise docu-

mentary evidence concerning the different manufac-

turing steps and system statuses. Raising the level of

integration and sharing additional data between par-

ticipants would provide much greater insight into exe-

cuted production processes and signiﬁcantly increase

the transparency within a VO.

On the other hand side, the issue of exposing

conﬁdential enterprise data to collaborating parties

should also be addressed when facilitating VOs. If

not adequately managed, malicious participants could

misuse exchanged sensitive data. Therefore, ex-

changed data must be protected by imposing strict au-

thorization rules that regulate whom and under what

circumstances may obtain it. Participants should also

be provided with a mechanism that enables conﬁgu-

ration of what production data should be shared with

other participants of a VO while preserving the conﬁ-

dentiality of private enterprise data.

By emphasizing data privacy and transparency as

key attributes of a solution that facilitates VOs, we

propose an environment that secures trustworthy con-

ditions for cooperation, i.e., participants could rely

on the system to ensure a certain degree of trust is

achieved and maintained within a VO. Our research

aims at assisting the VO cooperation within NHNs

by offering a novel methodological approach for sup-

porting the execution of collaborative production pro-

cesses. The approach allows VO participants to (i)

model collaborative production processes while pre-

serving the conﬁdentiality of private enterprise data

and (ii) conﬁgure what data should be shared be-

tween participants during the production execution.

Although our approach primarily aims to support pro-

duction processes, the concepts it introduces can be

applied to other VO domains as well.

On the implementation level, the approach will

be facilitated with the use of a DLT platform

and blockchain technology with smart contracts for

VO integration. DLT is a type of a distributed

database, while blockchain represents a distributed

data structure that implements DLT and comprises

cryptographically linked blocks that contain im-

mutable records of network transactions (Hileman

and Rauchs, 2017). Using a DLT platform would

provide VO participants with a mechanism for dis-

tributing production records that assures a shared con-

trol over the access to and evolution of data. Having

a shared control over data evolution would increase

transparency within a VO, as all stakeholders would

be aware of changes made to the shared data. Data

stored on a DLT platform are generated and validated

using smart contracts – computer programs whose ex-

ecution is guaranteed by system rules and for which

the outcome of execution is veriﬁable and auditable

by all network participants. Smart contracts have

the potential to improve coordination and veriﬁca-

tion within a VO by automatically verifying that the

production process actions are executed according to

the contracted production speciﬁcation. Data privacy

could be addressed by relying on a private, permis-

sioned, consortium-based DLT platform for storing

production records. Such platforms impose strict re-

strictions regarding ’read’ and ’write’ permissions on

the ledger (Wust and Gervais, 2018), which is critical

for protecting private enterprise data.

One downside of using DLT platforms is that

they usually provide low-level, general-purpose pro-

gramming languages for implementing smart con-

tracts. This is not always suitable for facilitating

VOs because a manual speciﬁcation of smart con-

tracts would reduce the capability of VOs to syn-

chronize and adapt their production in a timely man-

ner. Moreover, it would mean that process design-

ers, i.e., process and quality engineers responsible for

the production speciﬁcation, need to be proﬁcient in

these languages. This downside could be mitigated by

(i) providing process designers with a modeling lan-

guage that is based on concepts and notations they

are familiar with and already use in their domains,

and then (ii) relying on automatic generation of smart

contracts (France and Rumpe, 2007). Thus, our ap-

proach is centered around a DSML that provides col-

laborating parties with concepts required for describ-

ing Cross-Organizational Business Processes (CBPs)

(Berre et al., 2007) in a sufﬁciently detailed and un-

derstandable way to enable the execution of speciﬁed

processes. It is also facilitated by a software solu-

tion in which the MD principles and DSMLs are used

to (i) specify contracted CBPs and (ii) automatically

EI2N 2021 - IFAC/IFIP International Workshop on Enterprise Integration, Interoperability and Networking

198

generate smart contracts used for production monitor-

ing. The described approach enables an analysis of

records of events that occur during the production and

allows parties to derive conclusions and determine if

there are any discrepancies between negotiated and

executed process steps. The approach aims to create

trustworthy conditions for collaboration between VO

participants and allow them to be more competitive in

the market.

The presented work is structured as follows. Apart

from Introduction and Conclusion, the paper has three

sections. Section II examines the background and re-

lated work for the approach. It also discusses the nota-

tional aspects and execution signiﬁcance of a DSML

used to create collaborative production process mod-

els. Next, Section III provides an outline of the ap-

proach. Finally, in Section IV, we present the capa-

bilities of the CE-MultiProLan DSML used to model

collaborative production processes.

2 BACKGROUND AND RELATED

WORK

There have been various efforts to assist SMEs in

forming and operating NHNs for the collaborative

development and delivery of customized products

(Priego-Roche et al., 2016). The Net-Challenge

project (Carneiro et al., 2010; Carneiro et al., 2014)

proposed a methodology for supporting the creation

and operation of VOs. The methodology is struc-

tured in ﬁve main phases: (i) Build and develop a

Business Community, i.e., a phase of creating an en-

vironment that comprises a signiﬁcant number of or-

ganizations interested in joining VOs, (ii) Qualify, a

phase where data about potential partners is collected

and stored to be taken into consideration when a VO

is being formed, (iii) Form, a phase in which a VO

is created, and product concept and cost estimations

are deﬁned, (iv) Operate, a phase in which detailed

production plan is deﬁned, after which the produc-

tion is executed and monitored, and (v) Dissolve, a

phase in which performance evaluation is conducted,

after which a VO is dissolved. The methodology also

introduces three types of roles included in VO man-

agement: (i) the Broker role, representing an orga-

nization responsible for coordinating a VO; (ii) the

Core partner role, representing organizations that col-

laborate actively in the formation of the VO and the

deﬁnition of the product concept; and (iii) the Addi-

tional partner role, representing organizations that are

invited to provide quotations for speciﬁc operations or

materials and do not participate in the product design

phase. Based on the methodology, authors (Shamsuz-

zoha et al., 2016) created an innovative solution for

supporting collaboration between SMEs and real-time

information sharing during production execution. IoT

devices in production plants are used to collect data

about possible production process disruptions during

the execution of the Operate phase. This data is then

processed and stored in a centralized database. Stake-

holders can access the collected data through a web

app to identify levels of alerts and statuses of executed

process activities (e.g., machine breakdown, shortage

of raw materials).

Data about process disruptions collected in the

presented solution still represents only a fraction of

data generated during the Operate phase of the VO

lifecycle. As a result, records of events that occur dur-

ing production are primarily stored and maintained

within isolated IT systems of participants that exe-

cuted a speciﬁc production activity. Sharing addi-

tional data between participants would provide much

greater insight into executed production processes and

signiﬁcantly increase the transparency within a VO.

In addition, participants should be provided with a

mechanism that enables modeling of production pro-

cesses and conﬁguring what production data should

be shared with other participants of a VO while pre-

serving the conﬁdentiality of private enterprise data.

The feasibility of our approach is impacted heav-

ily by the interoperability concerns of participant’s

production systems. To enable interoperable event-

sharing conﬁguration and distribution of production

data, we utilized concepts identiﬁed in the ATHENA

Interoperability Framework (AIF) (Berre et al., 2007),

a methodological framework that enables collabora-

tive modeling and execution of CBPs. AIF facilitates

the enactment of CBPs by merging research areas

that support the development of interoperable enter-

prise solutions, most notably: (i) enterprise modeling

used to deﬁne interoperability requirements and sup-

port solution implementation, and (ii) frameworks for

implementation of interoperable platforms.

Production process modeling is an important re-

search topic (Xu, 2011), but it is still not ade-

quately covered with the existing studies (Petrasch

and Hentschke, 2016). Different notational aspects

of production process models, regarding their expres-

siveness and conﬁguration of what data should be

shared in a VO, are examined in Section 2.1. Our re-

search also promotes trust-building between partners

by relying on a DLT platform and smart contracts for

data distribution. Different aspects of DLT platforms

that have signiﬁcance for the enactment and integra-

tion of CBPs are addressed in Section 2.2.

A Novel Approach and a Language for Facilitating Collaborative Production Processes in Virtual Organizations Based on DLT Networks

199

2.1 Event-sharing Conﬁguration based

on Process Modeling

Several research challenges should be taken into con-

sideration when modeling processes that are collabo-

ratively executed in a VO. On the one hand side, mod-

eling of CBPs implies that a modeling language sup-

ports designers in describing production process spec-

iﬁcations in a sufﬁciently detailed and understandable

way to enable the execution of speciﬁed processes.

On the other hand side, these speciﬁcations should be

displayed to related parties through different process

interfaces that facilitate understanding of collabora-

tion in a VO while preserving the conﬁdentiality of

private, internal enterprise information. Thus, one of

the most signiﬁcant challenges for designing such lan-

guage is to devise concepts that connect private pro-

duction processes with openly exposed process inter-

faces and combine different representations of intra-

organizational processes at CBP level (Lippe et al.,

2005). Additionally, the modeling language should

allow users to model details needed on the execution

level, e.g., showing invoked smart contracts and ex-

ecuted transactions, while separating CBP modeling

from a speciﬁc deployment architecture.

The collaboration is based on a distributed process

model where parties manage their part of the over-

all production process. Three different process types

have been investigated and customized for use in col-

laborative production to allow disclosure of private

process data to VO participants: (i) private produc-

tion processes that represent internal processes exe-

cuted by an organization; (ii) interface processes used

to coordinate internal actions with activities of exter-

nal partners while concealing private data; and (iii)

CBPs used to describe how participants collaborate

to produce the end product (Lippe et al., 2005). In

Section 3.1, we describe how models of these pro-

cess types are utilized to provide a mechanism that

enables conﬁguring what production data should be

shared between VO participants and generate smart

contracts that enable production monitoring based on

the shared data.

Modeling of private production processes has

been an important topic of our previous research

(Vje

stica et al., 2021b). The research resulted in

an MD approach, and a DSML tool named Multi-

ProLan, which can be used to model production pro-

cesses suitable for automatic execution in smart facto-

ries (Vje

stica et al., 2021a). By relying on the possi-

bilities of the MultiProLan, process designers can be

focused only on process steps that must be executed

and need not worry about production logistics and re-

sources that will execute the process steps.

The modeling of private production processes in

MultiProLan is performed at two different levels of

detail to make modeling easier for process designers.

First, the approach provides Master-Level (MasL)

models at a lower level of details, used by process

designers to create process models independent of the

production facility in which the production will be ex-

ecuted. MasL models contain operation and inspec-

tion activities with their corresponding inputs and out-

puts, capabilities required to execute them, and simi-

lar. Second, at a higher level of details, the approach

provides Detail-Level (DetL) models, created by en-

riching MasL models with details about resources

available in the speciﬁc production facility in which

the production will take place. DetL models are ei-

ther created manually, by a process designer, or are

automatically generated by an Orchestrator, a system

that delegates instructions to different smart resources

in a smart factory (Pisari

c et al., ). The approach we

introduce in this paper extends the capabilities of the

MultiProLan tool by providing concepts required for

collaborative production planning and execution. We

named this extension as Collaborative Extension of

MultiProLan.

MultiProLan was selected as a basis for mod-

eling collaborative production processes in our ap-

proach because it was built to support process design-

ers in modeling execution-ready production processes

in multiple levels of detail. In addition, it was de-

vised in a way that makes it relatively easy to extend

it to support additional concepts required for support-

ing CBPs. Also, the language has already been tested

in an assembly use case that included collaboration

between human workers and robots in a production

setting. Our approach could be based on the use of

a general-purpose process modeling language, like

BPMN, for supporting the collaborative execution of

production processes. However, BPMN lacks the se-

mantics of production processes as it was mainly tai-

lored to model business processes. Hence, BPMN

is not adequate for modeling production processes

ready for automatic code generation and execution of

the code, especially when modeling products, capa-

bility constraints, parameters, and the material ﬂow

(Vje

stica et al., 2021b). Still, we could not ﬁnd a ﬁt-

ting modeling language that adequately facilitates the

modeling of execution-ready production processes.

2.2 DLT as an Execution Platform

An execution platform that facilitates horizontal inte-

gration should provide mechanisms that guarantee a

secure and transparent distribution of records to re-

lated parties to achieve a common understanding of

EI2N 2021 - IFAC/IFIP International Workshop on Enterprise Integration, Interoperability and Networking

200

these events. The architecture recommended by AIF

can be expanded to encourage the use of a DLT plat-

form and smart contracts for information sharing and

supporting trust-building between parties. To identify

technical requirements that should be taken into con-

sideration when developing an execution platform, we

relied on a list of Quality Attributes highlighted by the

network for Interoperability Development of Enter-

prise Applications and Software (IDEAS) (Chen and

Doumeingts, 2003). We selected the most important

attributes that have to be considered when developing

a platform for sharing data during the enactment of

CBPs: (i) security, i.e., mechanisms that platform of-

fers to protect private enterprise data; and (ii) scalabil-

ity and performance, i.e., the possibility of a platform

to process a large amount of data generated by smart

resources on the shop ﬂoor.

There have been several attempts at utilizing DLT

platforms and smart contracts for supporting VOs.

For example, in (O’Leary, 2019), the author suggests

that, because of their unique nature, VOs provide an

important potential setting for the use of blockchain-

like designs. However, the presented paper primarily

deals with integrating participants’ accounting sys-

tems rather than integrating their production pro-

cesses and production systems. Multiple existing so-

lutions consider using DLT platforms for monitoring

the enactment of CBPs (Weber et al., 2016) (Klinger

and Bodendorf, 2020). These solutions are based on

BPMN and rely on an MD approach to generate smart

contracts that facilitate the integration of collaborative

processes supported by a DLT network. In these solu-

tions, authors present tools that take business process

models as an input and generate smart contracts that

are then deployed on the Ethereum public DLT net-

work (Chowdhury et al., 2019). Described methods

have limitations regarding their usability in the col-

laborative production scenario. The use of a public

DLT network like Ethereum may not ﬁt the high data

security requirements of the collaborative production

domain. Instead, enterprise solutions that rely on a

private, consortium federated DLT network should be

used to protect highly sensitive corporate data. In ad-

dition, these solutions lack mechanisms for preserv-

ing sensitive corporate data when modeling and exe-

cuting collaborative processes.

Scalability and performance may also become a

concern with the use of a public DLT network, where

each transaction needs to be processed by every sin-

gle node in the network. For instance, Ethereum sup-

ports up to 15 transactions per second (tps), which

creates a severe bottleneck when supporting the ex-

ecution of production processes in collaborative pro-

duction, where machines involved in manufacturing

generate transactions at a much higher pace. To the

best of our knowledge, none of the existing solutions

consider high security, performance, and scalability

requirements in a uniﬁed way.

2.3 Summary

During the research, we found several solutions that

rely on an MD approach for generating smart con-

tracts that support the execution of collaborative

processes within a VO. However, these solutions

lack mechanisms for concealing sensitive corporate

data when modeling and executing collaborative pro-

cesses. In addition, these solutions are based on

BPMN, which is not adequate for modeling produc-

tion processes ready for automatic instruction gener-

ation. Since we could not ﬁnd a ﬁtting solution, we

decided to introduce a novel methodological approach

presented in this paper. The approach enables VO par-

ticipants to model and execute collaborative produc-

tion processes while preserving conﬁdential informa-

tion by carefully selecting what private data will be

disclosed to collaborating parties. Furthermore, the

proposed approach uniﬁes all collaborative produc-

tion process aspects, as presented in the rest of this

paper, and thus enables the speciﬁcation of produc-

tion process models used for automatic smart contract

generation and production process execution monitor-

ing.

3 AN OUTLINE OF THE

APPROACH

To promote trustworthy and transparent collaboration

between participants of a VO, we introduce a method-

ological approach in which CE-MultiProLan is used

to model collaborative production processes and con-

ﬁgure what data should be shared between partici-

pants during the execution of CBPs. A software so-

lution that utilizes MD principles is used to generate

smart contracts based on these models automatically.

Finally, generated smart contracts are stored in a DLT

network and used for production execution monitor-

ing and trustworthy distribution of production data

between VO participants. This section presents the

outline of the approach. The complete methodologi-

cal approach will be described in detail in our follow-

ing paper.

The approach is based on the Net-Challenge

methodology (Carneiro et al., 2010) but introduces

improvements regarding how CBPs are modeled and

executed in a VO, allowing for a higher level of inte-

gration between participants while preserving private

A Novel Approach and a Language for Facilitating Collaborative Production Processes in Virtual Organizations Based on DLT Networks

201

enterprise data. The expected advantages of applying

the presented approach for supporting the collabora-

tive production execution within a VO are: (i) a more

real-time insight into production status; (ii) improved

trust between participants as transparency within the

network is increased, and contract validations are au-

tomated and tamper-proof; and (iii) faster time to mar-

ket due to the automatic generation of smart con-

tracts. These advantages jointly create trustworthy

conditions for collaboration between SMEs involved

in a VO and allow them to be more competitive in

the market. The outline of the approach is depicted

in Fig. 1. The modeling of collaborative processes,

shown on the left-hand side of the ﬁgure, is described

in 3.1, while smart contract generation and process

monitoring, shown on the right-hand side of the ﬁg-

ure, is described in 3.2.

The approach we outline in this paper was ini-

tially proposed in our previous paper (Todorovic et al.,

2020), but has since been reﬁned to ﬁt the VO do-

main better. To evaluate the feasibility of our ap-

proach, we have developed a simple implementa-

tion of a prototype solution. Within this solution,

we created the CE-MultiProLan modeling tool. We

also created a code generator that generates smart

contracts based on process models deﬁned with CE-

MultiProLan. The generated smart contracts were de-

ployed to an established DLT network, from where

users can access shared records and monitor the pro-

duction execution. The use of CE-MultiProLan is

demonstrated in Section 4.

3.1 Modeling Collaborative Processes

During the Form phase of a VO lifecycle of Net-

Challenge methodology, participants are required to

develop a product concept collaboratively and de-

ﬁne a corresponding Engineering Bill-Of-Materials

(eBOM). eBOM represents a structure of a prod-

uct at the product design phase. It contains rela-

tionships between product’s materials, parts and sub-

assemblies. In this lifecycle phase, a corresponding

Bill-Of-Operations (BOO) also needs to be deﬁned.

The operations speciﬁed in BOO are then allocated to

VO participants. Deﬁned eBOM and BOO documents

are used as input to our approach.

In the Operate phase, as the ﬁrst step of our ap-

proach, depicted as 1 in Fig. 1, eBOM and BOO are

used by collaborating parties to specify a CBP Model

(CBPM). A CBPM is created to coordinate produc-

tion between participants and contains a sequence of

production activities (e.g., fabricating a part, assem-

bling a product) allocated to participants responsible

for executing them. Collaborating parties can also use

CBPM to specify roles of the included participants,

production milestones, i.e., critical points in produc-

tion used to determine the state of an activity (e.g.,

how much time each activity requires), and a due date

for an activity.

Next, in step 2, VO participants need to de-

sign their allocated individual production operations

from BOO. Participant’s process designers are re-

sponsible for designing a MasL Production Process

Model (MasL-PPM), which represents a private pro-

duction process speciﬁcation containing: (i) produc-

tion steps, (ii) capabilities required to execute each of

the steps, (iii) input and output products, i.e., trans-

formed resources like raw materials and components

from eBOM, (iv) constraints and (v) capability pa-

rameters. After MasL-PPM is created, it is utilized

for creating a DetL Production Process Model (DetL-

PPM) in step 3, and an Interface Production Process

Model (I-PPM) step 4.

DetL-PPM is an execution-ready process model

generated by enriching MasL-PPM with details about

IoT devices, i.e., Smart Resources available in a spe-

ciﬁc production facility that satisfy the requirements

deﬁned in MasL-PPM. DetL-PPM is either created

manually by the process designer or automatically

generated by the Orchestrator and is used to gener-

ate commands for orchestrating dedicated resources

and executing the speciﬁed production process. This

paper does not cover DetL-PPM in detail as process

execution is out of the scope of our approach, but its

details can be found in (Vje

stica et al., 2021b).

I-PPM represents a public interface created over

a private MasL-PPM that provides an insight into

how the responsible VO participant executes a spe-

ciﬁc value-adding CBPM operation. It is created as

a viewpoint over MasL-PPM and is deﬁned by pro-

cess designers manually, with some details present in

MasL-PPM anonymized or concealed to preserve en-

terprise data privacy. I-PPM is also used to conﬁg-

ure what data should be shared between collaborating

parties by specifying (i) what production steps from

MasL-PPM should be traced during production exe-

cution and (ii) what data should be persisted along-

side those step traces. All operations from CBPM for

which the execution should be monitored must refer

to a corresponding I-PPM. Data to be exchanged dur-

ing CBPM operation execution is agreed upon with a

responsible VO participant.

3.2 Smart Contract Generation and

Production Monitoring

Smart contracts that monitor the enactment of CBPs

and track the state of each activity in the collaborative

EI2N 2021 - IFAC/IFIP International Workshop on Enterprise Integration, Interoperability and Networking

202

Figure 1: The outline of the approach.

production process are generated in step 5 of the ap-

proach using the Smart Contract Generator (SC Gen-

erator) component. For smart contract generation to

be possible, SC Generator must implement an algo-

rithm that enables translation I-PPMs and CBPMs to

executable smart contract code. The algorithm that

we used represents a modiﬁed version of the algo-

rithm presented in (Weber et al., 2016). In addition,

we rely on a Smart Contract Meta-Store (SCMS), a

component built and maintained by a business com-

munity. SCMS is used to store (i) smart contract

templates, (ii) id values for all generated smart con-

tracts, and (iii) data about references between smart

contracts. SC Generator uses I-PPMs and templates

from SCMS to generate smart contracts to monitor

a production process executed by a single partici-

pant. Based on CBPMs and templates from SCMS,

SC Generator automatically generates smart contracts

to monitor the enactment of CBPs.

Once smart contracts are generated, in step 6 they

are stored on a DLT Network to which all parties in-

volved in a business network have access. The DLT

Network and smart contracts represent a basis for im-

plementing the Data Exchange Component (DEC).

Smart Resources can send records about production

execution events from the shop ﬂoor to the DEC, au-

tomatically triggering actions speciﬁed as a part of the

stored smart contracts. The role of smart contracts,

generated based on CBPMs and I-PPMs, is to moni-

tor these records and validate that the production exe-

cution is conducted according to the contracted spec-

iﬁcations. As a ﬁnal step of the approach, step 8, col-

laborating parties can oversee the state of production

and contract fulﬁllment by looking at the immutable

store using a set of data visualization tools.

When selecting a production-ready DLT plat-

form with the appropriate characteristics, we have

taken into consideration the quality attributes deﬁned

in Section 2.2. In (Chowdhury et al., 2019), au-

thors presented a comparative analysis of the exist-

ing production-ready DLT platforms. Based on their

ﬁndings, we selected Hyperledger Fabric as the most

appropriate platform for facilitating the collaborative

production domain. The requirements related to the

security quality attribute were addressed by selecting

a private, permissioned, consortium-based DLT plat-

form administered by a set of identiﬁed parties in-

volved in a business network operating under a gov-

ernance model that enforces a certain degree of trust

(Wust and Gervais, 2018). Private DLT networks im-

pose restrictions on ’read’ access to the ledger, i.e.,

who can access the network and see transactions. Fur-

thermore, permissioned DLT networks allow only a

selected set of parties to submit new transactions to

the distributed ledger. In addition, Hyperledger Fab-

ric introduces advanced mechanisms for

Scalability and performance concerns have also

been considered. Machines used in production can

generate numerous records that need to be processed

by a selected DLT network with low latency to en-

able a sufﬁciently reliable distribution of data between

participants. By relying on the identities of partici-

pants, Hyperledger Fabric can use a more traditional

Crash Fault-Tolerant (CFT) consensus protocol that is

suitable for scaling the transaction throughput in the

network. The use of CFT enables Hyperledger Fab-

ric to support throughput of more than 3,500 tps, and

it has also been shown to allow up to 20,000 tps in

certain setting(Gorenﬂo et al., 2019). Such through-

put can cover many different collaborative production

use-cases.

A Novel Approach and a Language for Facilitating Collaborative Production Processes in Virtual Organizations Based on DLT Networks

203

4 AN APPLICATION OF

CE-MULTIPROLAN ON A

COLLABORATION USE CASE

In this section, we present a use case that demon-

strates the application of CE-MultiProLan, our

DSML developed to model collaborative production

processes and conﬁgure what data should be shared

between VO participants during the production exe-

cution. We used the Ecore meta-meta-model, which

is a part of the Eclipse Modeling Framework (EMF)

(Steinberg et al., 2008), to create the abstract syntax

of CE-MultiProLan. Also, we used the Eclipse Sir-

ius framework (sir, ) to create the graphical concrete

syntax of CE-MultiProLan and to enable the simple

implementation of a prototype solution.

The use case presented in this section demon-

strates the use of CE-MultiProLan for describing the

production process for a decorative wooden wine box

with an engraved acrylic front. It was devised to cover

core concepts for the domain of collaborative pro-

duction process modeling, introduced below. Models

covered in this use case and the smart contracts gen-

erated based on these models are available

CBPM, displayed in Fig. 2, is used to de-

scribe how VO participants collaborate to produce

the end product jointly. The graphical syntax of CE-

MultiProLan that supports the creation of CBPMs in-

troduces a pool, depicted as a rectangle with a light-

Figure 2: CBPM for a decorative wine box production.

https://github.com/TodorovicNikola/CE-MPL

blue ﬁlling for visual grouping of elements related to

a single VO. Within a pool, each participant of a VO

is assigned a swimlane, depicted as a rectangle with

a white ﬁlling, where each swimlane contains pro-

cess steps executed by one participant of a process.

For example, the presented CBPM displays a single

VO named ACME Wine Inc, in which three different

organizations collaborate to produce a wooden wine

box: (i) the Winery, with the Broker (B) role in the

VO, (ii) the Woodworking Company that has a role

of a Core Partner (CP), and (iii) the Acrylic Engrav-

ing Company, with the Additional Partner (AP) role

within the VO. The introduction of pools and swim-

lanes was motivated by similar concepts that exist in

BPMN. Here, they are used for creating clear and

well-structured models that specify roles and respon-

sibilities for each VO participant.

The production is initiated when the Winery re-

ceives an order (Start). After the production ini-

tiation, the CP and the AP perform their allocated

production activities in parallel as their activities are

modeled between parallelism (PAR) gates. The CP

produces a batch of wooden boxes (Produce Wooden

Box). The production must meet the speciﬁed con-

tractual clauses, stating that a total of 500 pieces

should be produced until the speciﬁed deadline. Af-

ter the production of the wooden boxes is completed,

they are shipped to the Winery (Ship Wooden Box).

At the same time, the AP engraves a batch of acrylic

front covers with a speciﬁed pattern (Engrave Acrylic

Front Cover) and ships processed front covers to the

Winery (Ship Front Cover). Finally, the Winery packs

wine bottles in the produced boxes and inserts the

acrylic front covers (Pack Wine Bottle). Transporta-

tion steps are marked with a yellow arrow, while the

value-adding operation steps are marked with a blue

circle with a plus sign in it. The plus sign indi-

cates that a CBPM operation represents a high-level

process step composed of low-level steps speciﬁed

in the associated I-PPM. I-PPM provides an insight

into how the responsible VO participant executes the

CBPM operation to complete the collaborative oper-

ation. I-PPM also deﬁnes what data will be shared

between VO members for each CBP operation during

production execution.

Fig. 3 illustrates MasL-PPM for the Produce

Wooden Box operation, performed by the CP (left)

and corresponding I-PPM (right). Process designers

use MasL-PPM to deﬁne the operation and inspection

steps that need to be executed during production. The

presented MasL-PPM model is composed of six parts:

(i) the start process step; (ii) parallel process steps of

assembling left-bottom (L-B) and right-upper (R-U)

sides of a box, after which these two assembled sides

EI2N 2021 - IFAC/IFIP International Workshop on Enterprise Integration, Interoperability and Networking

204

Figure 3: MasL-PPM (left) and I-PPM (right) for the Produce Wooden Box CBPM step.

are assembled into a frame; (iii) hammering a back-

side panel into the frame; (iv) a manual inspection of

the box that veriﬁes that it conforms to the speciﬁed

constraints; (v) decision whether the box needs to be

stored or discarded, depending on results of the in-

spection process step; and (vi) the end process step.

As depicted, process step names can contain work in-

struction numbers, e.g., Discarding Defective Com-

ponent step in the model, or id values assigned to pro-

cess steps by an Enterprise Resource Planning (ERP)

tool, present in all other operation or inspection steps

in the model.

Input products, output products, and capabilities

of a process can be hidden from a diagram using a

+/- button at the top left corner of a process step so

that a process diagram could be more or less complex

depending on the designer’s needs. Due to the length

limitations of this paper, Fig 3 depicts these detailed

speciﬁcations for just two process steps, necessary for

explaining different concepts, while for the rest, they

are speciﬁed but not displayed.

The process step of assembling the L-B side rep-

resents an operation, depicted with a circle icon at

the left side of the process step name. It has two in-

put products, left and bottom sides of the frame, both

collected from a storage. The inverted triangle icon

at the left side of a product name indicates that an

input product should be collected from a storage or

that an output product should be placed in a storage.

Two input products are associated with two dimen-

sion constraints, width and height, that will be con-

sidered by the Orchestrator when it assigns a smart

resource that is able to pick the plank of these dimen-

sions. The same process step has the Assemble capa-

bility with parameters representing two wooden pins

with the space between them of 7mm, and a maxi-

mum time in which this step should be executed. The

output product of this process step is the assembled

L-B side, which will not be stored but will be used

by the following process step. Assembling the R-U

side is an equivalent process step to assembling the

L-B side process step. Both process steps can be ex-

ecuted in parallel, as they are modeled between two

PAR gates. The following process step, Frame As-

sembling, requires assembling the frame and has two

input products - L-B and R-U sides from the previous

two process steps. After the Frame Assembling pro-

cess step is ﬁnished, the back side is hammered into

the frame. Next, the Box Inspection process step is

performed to inspect if the box has any defects. Here,

a decision to store or discard the box should be made

depending on whether the box passes all checks or

not, e.g., if dimension deviations meet the box speci-

ﬁed values. Storing and discarding the box steps are

modeled between two decision gates (DEC). The pro-

cess is ﬁnished after it reaches the End process step.

A Novel Approach and a Language for Facilitating Collaborative Production Processes in Virtual Organizations Based on DLT Networks

205

I-PPM, which corresponds to the described MasL-

PPM, is shown on the right-hand side of Fig 3. It is

created by adapting details available in MasL-PPM

for displaying them to collaborating parties without

disclosing conﬁdential production details. Several

mechanisms have been introduced to protect the con-

ﬁdential information of participants. First, for each

of the elements in the model, e.g., process steps, in-

put and output products, capabilities, and constraints,

the process designer can choose whether or not to

expose them to the collaborating parties. Thus, el-

ements from the MasL-PPM that should not be ex-

posed are omitted from I-PPM. For example, input

products of the L-B and R-U side assembling pro-

cess steps are omitted from the I-PPM, as they are

regarded as conﬁdential. Similarly, conﬁdential capa-

bilities and constraints can also be hidden (e.g., space

between wooden pins for the L-B and R-U side as-

sembling process steps, or max pin distance deviation

in the Measure Dimensions process step). Second,

since process step names can sometimes disclose pri-

vate information, e.g., work instruction numbers and

ERP id values, aliases were introduced. By introduc-

ing aliases, process step labels in I-PPM can contain

customized values different from those displayed in

MasL-PPM, thus concealing private information.

I-PPM is also used to conﬁgure what data should

be shared with collaborating parties during produc-

tion execution. Process steps that should be traced

during production execution are displayed as double-

edged rectangles. In contrast, process steps that

should not be traced are shown as rectangles with a

single edge (e.g., Handle Defective Box process step).

The collaborating parties might decide not to trace a

speciﬁc process step if deemed irrelevant for the col-

laboration context. When process steps are traced, ad-

ditional data can be shared with collaborating parties.

Id values for exposed input and output products used

within process steps are automatically shared with

collaborating parties as they are signiﬁcant for prod-

uct provenance. Other details about input and output

products are displayed in the model only for docu-

mentation purposes. Capabilities and constraints also

specify additional data that should be shared with col-

laborating parties alongside process step traces. For

example, the Measure Dimensions process step has

the related Inspect capability containing disclosed in-

spections performed on every produced wooden box.

For constraints given in bold font, data will be shared

with collaborating parties, while constraints displayed

with a gray color are shown in the model only for doc-

umentation purposes, and data about those constraints

will not be shared. In addition, for each constraint,

parties can specify if shared data should be stored as

a Plain Text (PT), as a Hashed Value (HV), or an En-

crypted Value (EV).

The presented CBPM and I-PPM are suitable for

use in the collaboration scenario. CBPM is based on

concepts that allow collaborating parties to describe

how they collaborate to produce the end product. Fur-

thermore, I-PPM supports concepts that allow col-

laborating parties to coordinate internal actions with

activities of external partners while concealing pri-

vate data. For this to be possible, we have extended

the scope of concepts supported by the MultiProLan

DSML with those required for collaborative process

modeling. In addition, I-PPM was built as a view-

point over MasL-PPM, which makes it easier for pro-

cess designers to generate I-PPM based on the exist-

ing MasL-PPM. By relying on a clear separation of

collaborative process models, interface process mod-

els, and private process models, collaborating parties

can jointly plan production while preserving private

enterprise data.

5 CONCLUSION

This paper presents an overview of a novel method-

ological approach that promotes trustworthy and

traceable collaborative production execution within a

non-hierarchical VO. The approach is based on the

capabilities of the presented CE-MultiProLan DSML

used by process designers to (i) model collabora-

tive production processes while preserving the con-

ﬁdentiality of private enterprise data and (ii) con-

ﬁgure what data should be shared between partici-

pants during the execution of CBPs. Process mod-

els designed using CE-MultiProLan DSML are then

used as an input for a software solution that relies on

MD principles to generate smart contracts automat-

ically. Finally, generated smart contracts are stored

in a DLT network and used for production execu-

tion monitoring and trustworthy distribution of pro-

duction data between VO participants. The use of

CE-MultiProLan is demonstrated in this paper on a

use case that describes the production process for a

decorative wooden wine box with an engraved acrylic

front.

The core part of the MultiProLan language has

been tested by process designers on the shop ﬂoor

within a small-scale industrial production setup. It

has also been evaluated through a questionnaire, and

the evaluation results have been published (Vje

stica

et al., 2021a). The evaluation conclusion is that Mul-

tiProLan fulﬁlls all the following quality characteris-

tics: functional stability, usability, reliability, expres-

siveness, and productivity. The language extension

EI2N 2021 - IFAC/IFIP International Workshop on Enterprise Integration, Interoperability and Networking

206

that supports the collaborative execution of produc-

tion processes introduced in this paper will be sys-

tematically evaluated and tested on a case study com-

mon for VOs with a non-hierarchical structure. Fur-

thermore, we plan to improve the possibilities of CE-

MultiProLan for modeling collaborative production

processes by expanding the set of concepts available

on the interface process level. For example, advanced

concepts already present on the private process level,

like sub-processes and unordered steps, should also

be available on the interface level. The support for

the newly introduced concepts should also be imple-

mented in the software that generates smart contracts.

In addition, we plan to investigate the possibility of

utilizing enterprise modeling constructs deﬁned in the

newly introduced ISO standard for Enterprise Mod-

elling and Architecture (ISO 19440:2020, 2020). This

standard focuses on engineering and the integration of

manufacturing and related services in the enterprise.

We plan to analyze the possibility of using those con-

structs for production process modeling. Even though

the standard is rather new, there have already been at-

tempts to utilize it in the production modeling context.

For example, authors in (Wu et al., 2021) utilize it for

extracting object classes for a meta-model used to de-

scribe Cyber-Physical Production Systems(CPPS).

Even though our solution introduces a tamper-

proof and immutable way for storing production data,

the issue of when a malicious partner tries to submit

incorrect data about process execution persists. Al-

though the production process automation somewhat

mitigates this issue by decreasing the space for man-

ual data input, there is still a real possibility that a ma-

licious collaboration partner could try to submit the

wrong data. For this reason, we plan to investigate the

possibility of detecting such behavior with a compo-

nent that would perform smart contract execution log

analysis. As a part of this investigation, we published

a paper that expands further on the topic of ﬁnding

discrepancies between contracted and executed pro-

duction processes (Ivkovi

c and Lukovi

c, 2021).

The expected value of our approach for parties in-

volved in a VO is increased structural transparency

during the enactment of collaboration as contracts are

automated and tamper-proof. In addition, the ex-

pected scientiﬁc implication is a novel methodolog-

ical approach for the integration of VO participant’s

information systems that would create conditions for

trustworthy and traceable production execution.

ACKNOWLEDGEMENTS

This paper is supported by the Ministry of Education,

Science and Technological Development through the

project no. 451-03-68/2020-14/200156: “Innovative

scientiﬁc and artistic research from the FTS domain”.

REFERENCES

Eclipse sirius documentation. https://www.eclipse.

org/sirius/doc/. Accessed: 2021-07-29.

Berre, A. J., Elvesæter, B., Figay, N., Guglielmina, C.,

Johnsen, S. G., Karlsen, D., Knothe, T., and Lippe,

S. (2007). The athena interoperability framework. In

Enterprise Interoperability II, pages 569–580.

Carneiro, L., Shamsuzzoha, A., Almeida, R., Azevedo, A.,

Fornasiero, R., and Ferreira, P. S. (2014). Reference

model for collaborative manufacturing of customised

products: applications in the fashion industry. Pro-

duction Planning & Control, 25(13-14):1135–1155.

Carneiro, L. M., Almeida, R., Azevedo, A. L., Kankaanpaa,

T., and Shamsuzzoha, A. H. M. (2010). An Innovative

Framework Supporting SME Networks for Complex

Product Manufacturing. In Collaborative Networks

for a Sustainable World, volume 336.

Chen, D. and Doumeingts, G. (2003). European initia-

tives to develop interoperability of enterprise applica-

tions—basic concepts, framework and roadmap. An-

nual Reviews in Control, 27(2):153–162.

Chowdhury, M. J. M., Ferdous, M. S., Biswas, K., Chowd-

hury, N., Kayes, A. S. M., Alazab, M., and Wat-

ters, P. (2019). A Comparative Analysis of Dis-

tributed Ledger Technology Platforms. IEEE Access,

7:167930–167943. Conference Name: IEEE Access.

France, R. and Rumpe, B. (2007). Model-driven Devel-

opment of Complex Software: A Research Roadmap.

In Future of Software Engineering (FOSE ’07), pages

37–54.

Gorenﬂo, C., Lee, S., Golab, L., and Keshav, S. (2019).

FastFabric: Scaling Hyperledger Fabric to 20,000

Transactions per Second. arXiv:1901.00910 [cs].

arXiv: 1901.00910.

Hileman, G. and Rauchs, M. (2017). 2017 Global

Blockchain Benchmarking Study. SSRN Electronic

Journal.

ISO 19440:2020 (2020). Enterprise modelling and architec-

ture — constructs for enterprise modelling. Standard,

International Organization for Standardization.

Ivkovi

c, V. and Lukovi

c, I. (2021). An approach to vali-

dation of business-oriented smart contracts based on

process mining. Bellatreche L. et al. (eds) New Trends

in Database and Information Systems. ADBIS 2021.

Communications in Computer and Information Sci-

ence, 1450:303–309.

Klinger, P. and Bodendorf, F. (2020). Blockchain-based

Cross-Organizational Execution Framework for Dy-

namic Integration of Process Collaborations. In

WI2020 Zentrale Tracks, pages 893–908.

A Novel Approach and a Language for Facilitating Collaborative Production Processes in Virtual Organizations Based on DLT Networks

207

Lippe, S., Greiner, U., and Barros, A. (2005). A Survey

on State of the Art to Facilitate Modelling of Cross-

Organisational Business Processes. pages 7–22.

O’Leary, D. E. (2019). Some issues in blockchain for ac-

counting and the supply chain, with an application of

distributed databases to virtual organizations. Intel-

ligent Systems in Accounting, Finance and Manage-

ment, 26(3):137–149.

Petrasch, R. and Hentschke, R. (2016). Process modeling

for industry 4.0 applications: Towards an industry 4.0

process modeling language and method. In 2016 13th

International Joint Conference on Computer Science

and Software Engineering (JCSSE), pages 1–5.

Pisari

c, M., Dimitrieski, V., Vje

stica, M., Krajoski, G., and

Kapetina, M. Towards a ﬂexible smart factory with

a dynamic resource orchestration. Applied Sciences,

Special Issue Information and Communications Tech-

nology for Industry 4.0, MDPI, 11:7956.

Priego-Roche, L. M., Front, A., and Rieu, D. (2016). A

framework for virtual organization requirements. Re-

quirements Engineering, 21(4):439–460.

Priego-Roche, L.-M., Rieu, D., and Front, A. (2009). A

360° Vision for Virtual Organizations Characteriza-

tion and Modelling: Two Intentional Level Aspects.

In Software Services for e-Business and e-Society, vol-

ume 305.

Shamsuzzoha, A., Kankaanpaa, T., Carneiro, L. M.,

Almeida, R., Chiodi, A., and Fornasiero, R. (2013).

Dynamic and collaborative business networks in the

fashion industry. International Journal of Computer

Integrated Manufacturing, 26(1-2):125–139.

Shamsuzzoha, A., Toscano, C., Carneiro, L. M., Kumar, V.,

and Helo, P. (2016). ICT-based solution approach for

collaborative delivery of customised products. Pro-

duction Planning & Control, 27(4):280–298.

Steinberg, D., Budinsky, F., Paternostro, M., and Merks, E.

(2008). Eclipse Modeling Framework Second Edition.

Todorovic, N., Vjestica, M., Dimitrieski, V., Zaric, M.,

Todorovic, N., and Lukovic, I. (2020). Towards trust-

worthy horizontal integration in industry 4.0 based on

dlt networks. Proceedings, Federated Conference on

Computer Science and Information Systems (FedC-

SIS), 4th International Conference on Lean and Agile

Software Development (LASD 2020), 22:63–69.

Vje

stica, M., Dimitrieski, V., Pisari

c, M., Kordi

c, S., Risti

S., and Lukovi

c, I. (2021a). Multi-level production

process modeling language. Journal of Computer

Languages, Elsevier, 66:101053.

Vje

stica, M., Dimitrieski, V., Pisari

c, M., Kordi

c, S., Risti

S., and Lukovi

c, I. (2021b). Towards a Formal Speciﬁ-

cation of Production Processes Suitable for Automatic

Execution. Open Computer Science, 11(1):161–179.

Weber, I., Xu, X., Riveret, R., Governatori, G., Ponomarev,

A., and Mendling, J. (2016). Untrusted Business Pro-

cess Monitoring and Execution Using Blockchain. In

Business Process Management, volume 9850, pages

329–347.

Wu, X., Goepp, V., Siadat, A., and Vernadat, F. (2021). A

method for supporting the transformation of an exist-

ing production system with its integrated Enterprise

Information Systems (EISs) into a Cyber Physical

Production System (CPPS). Computers in Industry,

131:103483.

Wust, K. and Gervais, A. (2018). Do you Need a

Blockchain? In 2018 Crypto Valley Conference on

Blockchain Technology (CVCBT), pages 45–54.

Xu, L. D. (2011). Enterprise Systems: State-of-the-Art and

Future Trends. IEEE Transactions on Industrial In-

formatics, 7(4):630–640.

EI2N 2021 - IFAC/IFIP International Workshop on Enterprise Integration, Interoperability and Networking

208