Advances in mobile computing, Web services and in participatory sensing devices, along with the cost reduction of Internet of Things (IoT) technologies, have accelerated the development of a new generation of software systems and services including Cyber-Physical Systems (CPS). The term “cyber-physical systems” emerged around 2006. It was coined by Helen Gill at the National Science Foundation in the United States. Although it all about the frontier and seamless integration between the digital and physical worlds, diverse definitions have been proposed making the concept somehow confusing:
- Cyber physical systems are physical, biological, and engineered systems whose operations are monitored, and controlled by computational and digital components (software products, systems or services) that are tightly networked forming a system of software systems,
- Cyber physical systems are a composition of artifacts of the digital world that are “deeply embedded” into everyday physical component, possibly even into materials and hardware objects,
- Cyber-physical systems are characterized by a behavior that is a fully-integrated hybridization of computational and physical actions (Gill, 2006).
- Cyber-physical systems are “engineered systems that are built from, and depend upon, the seamless integration of computational algorithms and objects of the physical world” (NSF)
- Cyber physical systems are seen as platform for running smart services on interacting networks including physical and computational components (NIST)
- In cyber-physical systems, physical and software components are deeply intertwined, each operating on different spatial and temporal scales, exhibiting multiple and distinct behavioral modalities, and supporting interactions between humans, software systems and physical objects using a myriad of interfaces that change with context of use.
Most existing CPS today adopt the Internet of Things (IoT) and software as a Service (SasS) paradigm to provide innovative smart services for surveillance and critical infrastructure protection, environmental monitoring, management of vehicular traffic and transportation, control of manufacturing activities in industry, smart grid and energy systems, autonomous automobile and automatic avionics vehicles, medical and elderly assistance and monitoring as well as the socio-technical system of all these CPS.
One main characteristic due to the fast increase in the complexities and the operations of CPS involve the sensing, processing and storage of a huge amount of data. We argue that cyber physical are “big data-intensive systems”. Furthermore, as an eﬀect of this increasing penetration of these networked CPS in our environments and our lives is that our society is rapidly being “data centric”. Many of the devices we use, the networks through which they connect – not just the Internet but also mobile phones and embedded sensor networks – and the interactions we experience with these technologies (e.g., use of credit cards, driving on public highways, online shopping) have already generated considerable data. Data are produced by humans – whether volunteered via, e.g., for example social networks, or observed as with via online shopping or networked games – They are inferred and produced by others – not just the results of people-people interactions but, increasingly, by systems including CPS too.
This imposes fundamental design and human-centric validation of CPS in multiple aspects such as performance, sustainability (energy efficiency, among others), security, reliability, resilience, scalability as well as the required trade-offs between these attributes and those human factors such as usability, usefulness and privacy. Tackling these challenges necessitates innovative big data techniques for handling massive data in CPS including innovative human-data interactions techniques and ways for engaging users in processing highly interrelated data. Techniques are needed for big data visual analytics/mining and for the extraction of useful patterns that can be used for designing new services for controlling and monitoring physical objects and for the seamless integration with the digital world.
An important human barrier to operator-mediated CPS stems from the need to create interfaces (human-human, human-systems, and systems-systems) for the modeling, control, and adaptation of CPS driven by human experiences. For example multi-scale performance and functionality security needs to be integrated at human response time scales with appropriate compromise to balance between user satisfaction and system performance. Predictive patterns and measures to ensure congruent local context of use by one user with one service, so that incongruent actions by different people with the entire CPS are known and taken into account. How we shall express local context, make CPS and underlying services aware of it, and incorporate it with both the objectives of users and the behavior of system in its environment remain important open problems for CPS.
Traditional HCI and human-centered computing for cyber physical systems need to be upgraded. In CPS, the human can be seen as a physical part being connected with digital components while allowing human to interact with both physical and digital components. Humans can be also sensed using wearable and body-embedded sensors to control and customize their surrounding living space. This vision of the role and place of human in CPS might hope to break new ground of thinking systems – how do we successfully marry what people are good at with what computers are good at, in order to help people accomplish tasks – but rather by moving up a level and asking, for example, how we achieve certain systems “qualities” (reliability, security, scalability, safety, etc.) when humans are part of the system.
This vision of CPS requires the understanding human experiences beyond the current UX practices, i.e. humans with intention, purpose, and behavior. GPS and various kinds of sensor data are being available and integrated to mobile and Web services is what make this vision technologically feasible. For example, if an elderly person is about to get up, their cell phone might emit controls to his bed, shower, heating system, coffee machine., etc., Automatic control by humans sense and behavior of CPS will ensure that we no longer need to provide manual user interfaces in some cases, despite the need for potential intervention by human operators in the event of anomalous system behavior. For example, pilot/autopilot interactions increasingly appear in multiple CPS domains (e.g., skid auto-steering in automotive applications).
New science and theory of HCI design and user testing is needed to define safe yet usable human and cyber-based control, cyber-physical interlocks, protocols for interactions, and maintaining the human’s mental model and appropriate level experiences. Designers of human-CPS interactions need to consider additional issues such as: trade-offs between thinking (computation), talking (communications), and moving (control) in terms, for example, of power consumption (sustainability) vs. performance and security. It takes also new theories and models to capture the raw-data-to-trusted-knowledge dependency chain of processing. It is also valuable to feedback assessments of the created knowledge to the physical layer. Most current techniques do not act across all layers or operate as adaptively as will be required by open CPS. Mining of data streams in real time is significantly different than data analytics from static data. More than simple throughput is required, though that is a challenge in itself. We need to better understand how to keep and manipulate several visual representations of large chucks of data, some closer to the sampled data and some more abstract, so that we human can efficiently search, browse and comprehend the big data created by CPS. We also need history- and information-aware data repositories that allow us to retrieve meaningful patterns in a more direct/effective/targeted way.