Cyber-physical systems research focuses on knowledge and different engineer-ing principles integration to contribute to CPS science and technology develop-ment (Baheti & Gill, 2011). Khaitan and McCalley (2014) states that CPS has be-come a key area of research. This is not surprising because many researchers em-phasize the importance of the future CPS as they envision different possibilities and benefits of the future technology. CPS can offer formidable growth opportu-nities and novel services (Poovendran et al., 2011). Rajkumar (2012) suggests that CPS have transformational effects to interaction and control between humans and physical world. He continues that customer needs and competition pressures different industries to develop CPS, for example, factory automation, healthcare, and defense. In the future, design and development of different systems are pre-dicted to be CPS or influenced by CPS (Khaitan & McCalley, 2014). Monostori et al. (2016) discuss the potential and expectations of CPS as these systems can have major effects on different aspects of life, and the possibilities seem to be endless as CPS can be created for almost every industry. Expectations seem to have grown over the past decade, as well as the importance towards CPS research.
Broy, Cengarle & Geisberger (2012) discuss the vision of CPS in which open ubiquitous systems are able operate in different contexts through learning abili-ties and connection to other similar systems. They emphasize that these systems must be able to evolve as the full benefit from these systems can be achieved only
if CPS can adapt seamlessly and safely to various contexts, and prove to be de-pendable system. Conti et al. (2012) envision the future of CPS as they discuss smart devices and advanced technologies:
“By exploiting these devices and various technologies, information about physical re-ality (e.g., collected through sensor nodes) is seamlessly transferred into the cyber world where it is elaborated to adapt cyber applications and services to the physical context, and thus possibly modifying/adapting the physical world itself through ac-tuators” (Conti et al., 2012, p. 2).
Monostori et al. (2016) envision CPS through depicting the maturation of CPS into full potential systems (see figure 1). In this depiction, the system is evolving from basic setting to a sophisticated self-optimizing CPS. This model has five lev-els of maturity from basic to self-optimizing. Every level introduces new capabil-ities, such as information generation, information processing, information link-ing, and interacting CPS.
FIGURE 1 CPS maturity model (Monostori et al., 2016, p. 623)
Other visions related to CPS are customer related. Broy et al. (2012) reminds that CPS must be able to address different customer needs. Besides different customer needs, CPS should be developed as a discreet system that does not affect the physical world in a disruptive way. Beverungen et al. (2019) discuss that ideal CPS should be inconspicuous by blending in the environment, and should not bother users with its presence.
CPS originates from embedded systems which are defined as “a computer system within some mechanical or electrical system meant to perform dedicated specific functions with real-time computing constraints” (Monostori et al., 2016, p. 623). This defines CPS to some extent as well, but the definition do not
comprise the advanced complexity of cyber-physical systems. Broy et al. (2012) discuss CPS being an outcome of the development and integration of “systems with embedded software” and “global data networks like the internet with dis-tributed and interactive application systems” (p. 2). In addition to the depiction on the CPS origins, Broy et al. (2012) compares CPS to embedded systems as they state that CPS are “open sosio-technical systems that provide a range of new functionalities, services and features which go far beyond the current capabilities of embedded systems with controlled behavior”(p. 7). This comparison illumi-nates the superiority of CPS when considering the future possibilities and ad-vancements in technology.
CPS are high-confidence systems that consist of computational and physical capabilities which are integrated together to interact with physical environment (Baheti & Gill, 2011). Rajkumar (2012) discusses CPS in more detail as Rajkumar describes that CPS is a system which is monitored, coordinated, and controlled through communication capabilities. CPS are utilizing internet connection to ac-cess other data, and cloud computing for analyzing data (Monostori et al., 2016).
CPS definition by Khaitan and McCalley (2014) include also networking and con-text dependence, and they emphasize the close collaboration between the cyber and physical components that interacts continuously. The similar emphasis on collaboration is discussed by other authors as well. According to Monostori et al.
(2016), the intensive interaction between cyber and physical elements is essential, and they continue that it is important to understand that these elements are not unified but independent collaborators interacting with each other. Poovendran et al. (2011) also discuss the close connection between cyber and physical ele-ments as they describe this connection as closely controlled, tweakable, precise, and predictable.
One of the most important characteristics of CPS is the ability to interact with the physical world. Cyber capabilities are embedded to various elements to sense and actuate with physical world (Poovendran et al., 2011). Broy et al. (2012) describe these elements as they discuss that CPS consist of “powerful infrastruc-ture of sensors, actuators and communication networks” (p. 2). The interaction between CPS and the physical world set requirements for the operation of CPS as its operation should be dependable, safe, secure, efficient, and in real time (Raj-kumar, 2012). Because there are endless variations of environments in the physi-cal world, it is essential that CPS can adapt. CPS consist of systems and services which are dynamically adaptable to different contexts as CPS is autonomous and aware of itself, as well as aware of its own capabilities and the environment where it is located therefore being location independent (Broy et al., 2012).
As there are different contexts where CPS operate there are also differences in the characteristics of CPS. Khaitan and McCalley (2014) state that CPS differ with each other mainly having different characteristics, levels of operations, and what kind of applications CPS are used. They mention few common characteris-tics of CPS: cyber capabilities of physical elements, automation on high-level, net-working, integration, and reconfiguration. Broy and Schmidt (2014) discuss on the technical characteristics as they mention the most usual ones that are
included in CPS are actuators controlled by computing, network connectivity, sensors enabling environmental awareness, actuators interacting physical envi-ronment, distributed systems, and real-time operations. Beverungen et al. (2019) bring up similar notions as they mention four important capabilities of CPS: mon-itoring, control, optimization, and autonomy.
CPS have started to surface in different industries, for example, transporta-tion, communicatransporta-tion, entertainment, and manufacturing (Broy & Schmidt, 2014).
Khaitan and McCalley (2014) have listed different industries that have been linked to CPS in various articles, and the list consists of 18 different industries.
Clearly, CPS are objects of interest within many industries and this interest will probably expand to many other industries as well in the coming years like Mo-nostori et al. (2016) suggest that “the potential application fields are almost end-less” (p. 624).
Usually CPS are complex, but applications can also be simpler as Broy and Schmidt (2014) state that these systems are not limited to be complex and expen-sive. They give an example that the size of CPS can vary from handheld devices to factories. Other examples found from the literature include many different ap-plications, for example, modern vehicles, medical devices, robotic systems, fac-tory automation, and critical infrastructures. According to Khaitan and McCalley (2014) and Broy et al. (2012) modern vehicles are CPS because they consist of many different complex systems, for example, enhanced displays, motion and energy consumption systems, intelligent parking system, and navigation.
Beverungen et al. (2019) present a more recent example as they discuss on mod-ern washing machines that have different sensors for adjusting the operation of the product, for example, water and detergent consumption is adjusted accord-ing to the weight and dirtiness of laundry. They added that these products are connected to their manufacturer who can use the collected data of many products to fine tune the operation of all similar products. These complex activities require different technical features.
CPS consists of sensors, actuators, and computational and networking ca-pabilities as discussed previously. This combination enables the operation of CPS.
The computational and networking capabilities provides abilities for CPS to op-erate and adapt autonomously in different situations and environments as they can store and process data locally, and exchange data with other actors in their network (Beverungen et al., 2019). The processing and communication capabili-ties are orchestrating the operation of the product, but sensors and actuators are vital components in achieving the intended functions. As the sensors and actua-tors enable the interaction between the cyber and the physical world, CPS must coordinate how these heterogenous systems (sensors and actuators) including computational devices interact with each other (Khaitan & McCalley, 2014). Dif-ferent sensors enables CPS to examine the surroundings of it, and the insides of the physical product of CPS as well. CPS are active objects in the physical world as they collect physical data adjacent of themselves, and use the actuators to af-fect their surroundings, so CPS can become aware of the physical environment,
and be an actor in it (Beverungen et al., 2019). All this complexity increases the difficulty to build these systems.
Developing and building these complex systems pose new challenges. One major challenge relates to the novelty of CPS as Baheti and Gill (2011) describe that design and development of CPS is not supported by existing engineering and science knowledge. They continue that the challenge lies in building system and control methodologies which would operate as a platform for designing and operating CPS. Broy and Schmidt (2014) discuss the same challenge as they state that traditional design constraints do not apply to CPS, and knowledge for de-veloping CPS must be sought from multiple disciplines. They also discuss that developing CPS brings up requirements that are on a totally different scale than before, and there exist challenges related to the manufacturing, network integra-tion, and maintenance of CPS. Another challenge relates to the interaction of the physical and cyber world. Intensive pairing of these two fundamentally different worlds is extremely complex task (Poovendran et al., 2011). Accurate and appro-priately timed communication is essential but also challenging as CPS need seamless interaction between the cyber and physical world (Khaitan & McCalley, 2014). This complexity is also increased by continuously advancing technology.
When using more advanced components and technology for data processing, communication, sensors, and actuators, major challenges will emerge (Baheti &
Gill, 2011). Despite the different challenges, service orientation will increase with CPS and this will lead to new services.
CPS literature do not often mention services, but services are a part of CPS at some extent. Broy et al. (2012) describe that ”CPS are open thus dynamically adaptable systems and services” (p. 3). Also, Drath and Horch (2014) explain that there are three levels included in CPS: physical objects, data models, and services.
They specify that these services are based on the data that is available to the CPS, and CPS can also utilize third-party services, such as weather, calendar, geoloca-tion, historical data and payment solutions. Based on the literature, it seems that services are merely add-ons to CPS as these are basically smart products, such as smartphones. When considering the technological perspective, smart products are similar to CPS (Rizvi & Chew, 2018). Smartphones have different services, for example, smartphones are updated regularly and application markets provide new features in a form of applications. In the end, smartphones are products that customers purchase, and most of the services are additional, although some of the services can be critical for the smartphone operation.
Rizvi and Chew (2018) discuss cybernized services by using the concept cyber-physical product-service systems (CPSS), and they describe that CPSS is the combination of cyber-physical features and product-service systems. As they compare CPS to CPSS, they suggest that CPSS is much broader concept than CPS because it includes the aspects of service, value, actor network, and environment.
The complexity increases when service components, new stakeholders, new kind of interactions, and new disciplines are involved in the mix (Wiesner et al., 2017).