Industry 4.0 which is also referred to as the fourth industrial revolution aims at mass customized production with low-cost and shorter production life-cycle, by digitalizing and automating the manufacturing process. The Automation Systems (AS) used in the manufacturing process should be exible in order to achieve this goal. But, lack of interoperability between AS devices from di erent domains makes it harder to achieve this goal. In this study, we employ Web of Things and Semantic Web Technologies to address the cross-domain interoperability problems in AS. Then, we propose an approach to engineer and con gure an AS with minimum e ort; and show the preliminary results as a demonstration on a process automation workstation to show the feasibility of our approach.
Today, the AS used in manufacturing are increasingly becoming part of
Internet of Things (IoT). IoT is the system where the physical devices are
embedded into electronic systems which can connect to the internet and can be
discovered, monitored, controlled and interacted with each other over various
network interfaces. But, IoT lacks an universal application protocol which
hinders the integration of devices from various manufacturers into a single
application [
The main contribution of this project is to: create transparency of AS infrastructure, develop a low-e ort approach to engineer and con gure the interactions between AS devices in order to ful ll a goal speci ed by a manufacturing task. We also develop a low-e ort approach to re-engineer the AS devices to update their functionality, whenever it is required by a manufacturing process. We will make this semantic-based approach easily usable for an AS engineer with little knowledge of the SWT. 2
Web of Things: Initial standards are being developed for the Web of Things by the W3C WoT working group4 to address the concerns regarding interoperability between IoT platforms. The standard provides an interface for a physical thing called Thing Description5 (TD). TD describes a Thing in terms of its interactions and meta-data.
Industry 4.0: The Reference Architectural Model [
Semantic Web Technologies (SWT): provides standardized knowledge
representation formalisms such as RDF6, RDF Schema7 and OWL8 and query
language over the semantic data, called SPARQL9. There exists a number of
semantic models for WoT which provide domain-dependent and domain -
independent vocabularies. Few of them are: (1) Semantic Sensor Networks ontology
(SSN) [
The state-of-the-art engineering of AS devices is typically done using Software Engineering approach based on model-driven design. There are ve phases in the engineering of an embedded AS device. They are: 1. Design phase 2. Development phase 3. Engineering phase 4. Commissioning phase and 5. Operating phase. According to the model-driven design, an engineer speci es a eld function or a data point in a model in the Design phase. The code generation is run to produce a skeleton of a service that is supposed to implement the function or data point in the Development phase. Finally, in the Engineering phase, the engineer implements the service skeleton, deploys it in a service run-time (that is embedded in a device) in the Commissioning phase and starts it. The current way of engineering of embedded AS devices is time consuming and demands a lot of human e ort for implementation of the application and deployment. This method is not suitable for ever-changing nature of mass customized production. In contrast to this method, we propose a low-e ort engineering method which involves minimum human intervention in Section 4. 3
We have set the following research goals to achieve exibility in an AS for mass customized production: { Facilitate an AS infrastructure to be transparent in order to enable rapid application development. { Develop an end-to-end engineering approach to establish and con gure the interactions between the AS devices and re-engineer an AS to install new functionality on it with low-e ort and minimum human intervention. { Make the proposed semantic-based engineering approach easily usable for an
AS engineer with little knowledge of SWT.
7 https://www.w3.org/TR/rdf-schema/ 8 https://www.w3.org/OWL/ 9 https://www.w3.org/TR/rdf-sparql-query/ 10 https://www.eclass.eu/
We identi ed the following research challenges that should be addressed in order to achieve the goals mentioned above: { How to make an AS infrastructure transparent for rapid application development? { The AS are complex, there exists large number of functionalities and interconnections on an AS which should be taken into consideration during the engineering process. How to discover the functionalities on a complex AS? { How to engineer, con gure and re-engineer an AS with low-e ort, minimum human intervention? { We believe that SWT can tackle the above challenges. But, to make the proposed semantic-based approach usable for AS engineers with little knowledge of SWT is a challenge by itself.
The next section presents our approach to achieve the above mentioned research goals by addressing these challenges. 4
In this project, we develop a semantic-based approach for engineering and
conguring a WoT-enabled AS. Employing SWT in Web of Things is called SWoT
[
Traperency of the AS infrasturture: We model the WoT-enabled AS,
their services and applications semantically using W3C WoT Thing
Description (TD). A TD is semantically annotated by re-using the existing
domainindependent and domain-dependent models to enable cross-domain inter-
operability [
Low-e ort Engineering: In a manufacturing process a task is usually performed by letting the AS devices interact in a certain manner. It would be bene cial to store this composition of interactions as a template for a later use; In such a way that, a template can be e ectively discovered and extended with other templates. In order to achieve this, we have come up with a concept of 11 http://mqtt.org/documentation "Template based engineering" where interactions between certain devices are stored as an Engineering Template (ET). We develop a light-weight semantic model to describe an ET, which can be stored and discovered from a semantic repository as shown in Figure 1. In order to engineer an AS, an engineer discovers an ET from the repository and instantiates it with the matching devices on the AS, then the interactions between the devices is established by implementing and con guring the interactions using scripts and existing Web standard methods.
Low-e ort Re-engineering: In the mass customized production, the devices need to constantly update their functionality to meet the customer demands. In order to update new functionality easily, we have come up with a concept called Apps. An App is a semantic model which de nes an added-value functionality. Apps are stored in the semantic repository as shown in Figure 1, which enables discovery and re-use of the Apps. An App model describes the functionality of the App; and requirements that should be ful lled by a device to install the App. The App and the device semantic models enable automated compatibility check in the commissioning phase, to ensure that the device has the capability to run the App. Therefore, it minimizes the human intervention during re-engineering. Moreover, the App model is directly run-able on the device using device run-time, which minimizes the e ort during the engineering process.
Easy-to-use Engineering Tool: Our semantic-based approach for end-toend engineering could be complex to use for an AS engineer with little knowledge of SWT. Therefore, in oder to overcome this limitation, we develop an graphical tool for our engineering approach as shown in Figure 1. The tool should support an AS engineer in all phases of engineering. It should have the ability to guide an engineer to model TDs, ETs and Apps; and do semantic-based discovery on the semantic repository to discover the Apps, ETs and TDs and download them to the tool (see Figure 1). The tool should have the ability to communicate with an AS through a RESTful API to discover the functionalities on AS, install Apps and scripts on a device; or instantiate, implement and con gure an ET to engineer an AS. 4.1
Basic Building Blocks The devices under our consideration are resource-constrained devices which have limited memory and processing power, PC-based semantic reasoning techniques are not feasible on these devices [18]. Thus, in this project we use an embedded micro reasoner which can run on the resource-constrained devices (see Figure 1) with Unix / Linux platforms. The micro reasoner consists of two parts: a micro event processing engine implemented in C, which is based on the work from [19]. Event rules are used to do Complex Event Processing (CEP) of the events from the AS devices. The rules can be added/deleted easily to the engine over a RESTful API. The event rules and scripts can be directly deployed on the devices from the engineering tool as shown in Figure 1. The second part is a datalog reasoner12 which provides datalog reasoning on the device. The micro reasoner is embedded in the edge device which in our case is SIMATIC IOT204013. An edge device is embedded on an AS which acts as a gateway between the AS and the cloud. In some cases, the micro-reasoner can also run on the Class 2 resource-constrained devices [20] as shown in Figure 1.
Discovery of Semantically annotate
App & ET Register Semantic Models Download Download
Script Library Add/Update/ Delete/Query
Semantic-enabled end-to-end engineering tool Discover matching devices Deploys Apps, event rules, scripts
Edge Device
Micro reasoner
Micro Event Engine CPS Embedded Device
Thing Description ...
... We performed a feasibility test of our methodology presented above for the lowe ort re-engineering of an AS device to install new Apps on an AS with minimum e ort.
Demo setup: The implementation is demonstrated on the FESTO14 process automation workstation shown in Figure 2. The workstation is equipped with a few edge devices15 with micro reasoner running on them. All the sensors and the actuators on the workstation are embedded with NodeMCUs16 and connected to the edge devices. There is a binary oat sensor on Tank 1, which detects over ow in the tank. There exists a pneumatic valve which takes Boolean value as input and turns on the valve to pump out the liquid from Tank 1. The workstation is 12 http://www.ccs.neu.edu/home/ramsdell/tools/datalog/datalog.html 13 http://docs-europe.electrocomponents.com/webdocs/1536/0900766b815365c3.pdf 14 http://www.festo-didactic.com/int-en/learning-systems/processautomation/compact-workstation/mps-pa-compact-workstation-with-level, owrate,pressure-and-temperature-controlled-systems.htm 15 http://docs-europe.electrocomponents.com/webdocs/1536/0900766b815365c3.pdf 16 https://nodemcu.readthedocs.io/en/master/ engineered such that when over ow occurs in Tank 1 then the pneumatic valve will turn on to ensure over ow protection on Tank 1.
Use case: Our use-case was to ensure over ow protection on Tank 1. Imagine a situation where the oat sensor is malfunctioning then, the whole process will be disturbed. In order to avoid the machine downtime, we re-engineered the AS by installing an App on it which uses the liquid level values from the ultrasonic sensor (which measures the level of liquid in the tank) on Tank 1 as shown in Figure 2 to detect the over ow of the tank.
Demo steps: We implemented the proposed semantic-based re-engineering approach and corresponding features in the engineering tool. The tool provides an interface to the semantic repository to discover Apps, TDs and download them to the tool. The tool connects to the edge devices on the AS (through a RESTful API), and does distributed discovery for the devices and automated compatibility check between the App and the device, locally on the edge device. We installed an App on the discovered edge device which ful lls the App requirements. The App reads the liquid level values from the ultrasonic sensor and detects the tank over ow. Thus, the AS is re-engineered with low-e ort.
Result: We measured the time taken for distributed discovery and automated compatibility check. We tested the integration of engineering tool with semantic repository and its interaction with the AS. It proved that our approach is feasible to be applied on real-world use cases.
Tank 1
Ultrasonic sensor Edge devices
Pneumatic valve Float sensor
Tank 2 NodeMCU Embedded Ultrasonic sensor
Fig. 2: Festo MPS Process Automation Workstation
We demonstrate the engineering of an AS using ETs and re-engineering of an AS device on the following use cases: { Plug and Play use case: When a new device is introduced into the workstation, we will demonstrate how our approach can be used to re-con gure the an AS easily using engineering ETs with minimum human e ort, to adopt the new device into the system. After the feasibility test, we will also evaluate our approach on a large-scale industrial manufacturing unit. { Re-engineering use case: In addition to the demo on the FESTO workstation, we will demonstrate our approach for low-e ort re-engineering on a largescale industrial manufacturing unit.
We will perform quantitative analysis to check the (1) time taken for distributed discovery and automated compatibility check to nd matching devices, (2) time taken to implement and con gure an ET. We will perform a qualitative analysis as follows: we will invite the engineers of the AS with little knowledge of SWT to test the engineering tool: We will give the engineers a task to engineer an AS. (1) measure the time taken by them to do the engineering, (2) check the quality of their engineering using our approach, (3) give a questionnaire to the AS engineers and get their feedback regarding the pros and cons of the tool. 7
In this project, we proposed a semantic-based approach for low-e ort engineering and con guration and re-engineering of a WoT-enabled AS with minimum human intervention. This approach makes an AS exible for mass customized production. We demonstrated our implementation of re-engineering approach on a FESTO process automation workstation to test the feasibility of our approach. The following are the future steps in this project: (1) as a rst step, we develop an approach of engineering the AS devices with ETs, do distributed automated compatibility checks between the AS devices, instantiate an ET, implement and con gure the interactions between the AS devices with minimum e ort. (2) second, we work further on the engineering tool to design, engineer and implement ETs. (3) at last we do quantitative and qualitative analysis of our approach according the evaluation plan mentioned in Section 6. [15] Krotzsch, M., Rudolph, S., Hitzler, P.: Description Logic Rules. In: ECAI.
(2008) [16] Wahlster, W.: Semantic technologies for mass customization. In: In Towards the Internet of Services: The THESEUS Research Program. Springer (2014) 3{13 [17] Thoma, M., Braun, T., Magerkurth, C., Antonescu, A.: Managing things and services with semantics: A survey. In: In Network Operations and Management Symposium (NOMS). (2014) 1{5 [18] Christian, S., Rene, S.: Rule-based OWL reasoning for speci c embedded devices. In: The Semantic Web { ISWC 2011. Springer Berlin Heidelberg (2011) 237{252 [19] Anicic, D., Rudolph, S., Fodor, P., Stojanovic, N.: Stream reasoning and complex event processing in etalis. Semantic Web 3 (2012) 397{407 [20] Bormann, C., Ersue, Mand Keranen, A.: Terminology for constrained-node networks. RFC 7228 (2014)