Page 32 - Kaleidoscope Academic Conference Proceedings 2020
P. 32
model properties (i.e., data entries) and data sources (e.g., Due to their advantages with respect to flexibility and
analog inputs, bus telegrams) can be configured as well, as scalability, Service-oriented Architectures have major
illustrated in Figure 5. In consequence, the integration of potential in the production industry. They will become an
legacy devices into Industrie 4.0 environments is important architecture pattern for the realization of large-
significantly sped up. In general, providing off-the-shelf scale, flexible next-generation production systems. Today
solutions that can be commissioned quickly – ideally with no however, experiences and guidelines regarding SoA in
programming knowledge needed – has proven beneficial for industry are limited. It will take more experiences to
enabling a quick transition to SoA in manufacturing. incorporate the SoA principle into the automation industry.
4.3 Levels of SOA 5. RELATED WORK
Our experiences in different projects have shown that for This section provides a brief overview of existing
different partners, different degrees of the implementation of architecture concepts and SoA approaches for Industrie 4.0.
SoA for manufacturing are needed. For some partners, the Many existing approaches are not tailored to the production
primary goal was to get an overview of the overall domain. Instead, they target the Internet of Things (IoT),
manufacturing process, to collect data, and to display data in with production systems being one application domain. A
a comprehensible way. Here, implementing SoA only for comprehensive survey of IoT middleware and the challenges
data acquisition has already provided a great benefit. New it needs to address was conducted by Razaaque et al. in [10].
data sources can be easily integrated into the plant and the Research about middleware addressing Industry 4.0 is
dashboard. By utilizing AAS and sub-models, quick presented in a broad survey by Lu in [11]. A multitude of IoT
integration of devices from the same class (e.g., mills) was middleware platforms are offered by many different vendors.
made possible. AAS sub-models were defined once and
instantiated multiple times, once per device of this class. This Choosing the right platform for an individual business
common abstraction enabled quick integration into the application is difficult and requires effort. Nevertheless, the
dashboard. Beyond that, using SoA for the control of potential of using a modern communication platform is very
manufacturing processes is not desired by all partners. high in comparison to conventional plant automation. IoT
However, by instantiating the concept of SoA and the AAS, platforms can be differentiated in terms of their focus on
the foundation for a fully implemented service-based ambient data collection and analytics, while real-time
production has already been laid. Thus, companies can reactive applications provide features for changeability and
iteratively implement SoA in their plants without the need mass customization [12]. However, there is a lack of open
for a big-bang migration to a service-based production. IoT middleware that does not try to enforce the business
Instead, small iterations can be carried out, each resulting in model of its vendor [13], which limits the flexibility and
a usable SoA system. This illustrates that SoA is an freedom of the users of these platforms. The high upfront
architecture pattern rather than a fixed end-to-end solution. cost of adopting an IoT middleware platform quenches the
It also illustrates the flexibility that even modern production motivation in small and medium-sized manufacturing
systems can obtain by implementing SoA principles companies (SME). Open-source implementations like Kaa
gradually into production processes. (https://www.kaaproject.org/) or Eclipse BaSyx
(https://www.eclipse.org/basyx/) are readily accepted by
4.4 SoA issues SMEs due to their low upfront and operational cost.
The introduction of SoA into automation environments has 6. CONCLUSION AND FUTURE WORK
clear benefits, but also leads to new challenges. SoA is an
architecture paradigm that is fundamentally different from In this paper, we provided an overview of the application of
the cycle-based software architecture concepts being used the Service-oriented Architecture (SoA) principle to
today in automation technology. Changing these systems to production systems. SoA is not a single solution but a pattern
SoA requires fundamental changes in the software structure for the system architecture of production systems that
as well as additional software components (for example, a improves flexibility in production processes. The integration
service registry) that need to be deployed. Another challenge with additional Industrie 4.0 techniques, such as digital twins,
are the timing requirements in the automation domain. enables further benefits, such as end-to-end communication
Classic SoA systems have only soft real-time requirements and unified interfaces for applications and dashboards. The
or no real-time requirements at all. Production systems SoA principle, however, needs to be implemented through a
commonly have hard real-time requirements. Transporting software framework. We gave a survey of our experiences
heavy workpieces over a multi-segment conveyor belt implementing SoA in the automation domain. These
requires timely interaction of adjacent segments when the implementations were performed in real production
workpiece passes the segment border in order to prevent environments and yielded immediate benefits for the users.
physical damage in engines. Fulfilling these requirements During our activities, we noticed that the major challenge of
calls for new design approaches and real-time existing production environments are grown structures with
communication to address these aspects. These issues are proprietary interfaces that often prevent the users from
usually solved by applying the existing cycle-based field bus implementing novel approaches and ideas. The SoA
communication for a few isolated, critical control functions. principle gives them the opportunity to implement modern
– xxviii –