Cyber-Physical Systems (CPS) Explained

Cyber-Physical Systems refer to a system that models, automates and controls the mechanism of a physical system in a digital environment.

This is an area of significant growth: the global market for Cyber-Physical Systems (CPS) is expected to grow from around $87 billion in 2022 to over $137 billion by the year 2028 at a CAGR of 7.9%. 

So, what exactly are cyber-physical systems? Let’s take a look.

What is a cyber-physical system?

The U.S. National Science Foundation defines a cyber-physical system as:

A system that “integrate[s] sensing, computation, control and networking into physical objects and infrastructure, connecting them to the Internet and to each other.

In this system, both the physical and digital behaviors are deeply intertwined. A CPS allows users to replicate attributes of the physical system in a digital world.

Then, the dynamic behavior of a physical system across spatial and temporal domains is captured by software algorithms and then rendered in a consumable and intuitive digital user interface (UI).

CPS in Industry 4.0

Cyber-physical systems are integral to the Industry 4.0 movement, the fourth industrial revolution that is driven by hyper automation intelligence. It embeds intelligence and cognitive computing capabilities into the design and simulation process of a physical system.

These systems may involve complex operations, such as robots involved in precision manufacturing of nanodevices. By developing a digital twin of the physical instruments and processes, engineers can simulate changes and control operations from a centralized and unified interface.

Use cases of CPS

The applications of cyber physical systems are almost boundless. Today, they’re in use in healthcare and manufacturing industries all the way through to automotive, civil and energy industries. Some of the common use cases of CPS include:

• The architectural plans of infrastructure and buildings
• IoT devices
• Robotics
• Manufacturing
• Smart cities and smart grids

IoT vs CPS: is there a difference?

Yes, at first glance, cyber-physical systems might seem similar to the Internet of Things (IoT). They’re technically not the same. Remember that the internet is “simply a mechanism for transmitting information”.

So, making smarter products fundamentally different or better is not the internet as the messenger — it’s the way we design the “things”. Here’s how Vanderbilt University, in the U.S., clarifies the differences:

• IoT is the technology that enables the interconnection of all types of devices through the internet to exchange data, optimize actuators and monitor devices in order to generate results.
• Cyber-physical systems are comprised of computation and control components that are tightly bound with physical processes. This is the foundation for IoT that can bringing about advanced efficiency and connectivity of devices, systems and services in countless domains.

(Related reading: internet of medical things & IIoT, industrial Iot.)

Features of cyber-physical systems

Now let’s turn to the key features of cyber-physical systems:

Data-driven

Data is collected from cross-domain sensors and IoT devices. Then, you’ll develop an end-to-end data pipeline and data management program. This must be able to process semi-structured and unstructured sensor data in ways that are efficient, secure and reliable.

Embedded mobile sensing and data fusion

Small mobile sensors are embedded into physical objects. Data collected from a network of data sources is integrated together to produce consumable data with the necessary contextual knowledge and insights derived from network-wide information sources.

Adaptable and trainable models

After the data fusion process, AI models train on real-time information. The models may:

• Be trained using real-time or recurring information.
• Dynamically update according to the new information available.

This enables users to develop a correct cyber-physical systems model considering the dynamic states and diverse future projections of the information produced in the physical world.

(Related reading: adaptive AI & what generative AI means for security.)

Simulation- based design

The system design is run through exhaustive simulations to model the dynamics and produce an accurate, real-time feedback of the physical design characteristics. 

(Real-time feedback of systems is enabled by observability.)

Autonomous

The subsystems and cooperative components — both hardware and software — are designed for autonomy in three ways:

• Intelligence
• Accuracy
• Responsiveness

These characteristics allow the cyber physical system models to account for emergent dynamics and behavior of the physical systems without any human intervention or manual process control.

Continuous and real-time communication

IoT sensors and networking devices log information continuously via standardized communication protocols and API connectivity, as well as open-source software components.

Enable agnostic work: the digital system design is platform-independent and uses standardized communication middleware.

Scalability

A scalable data platform (such as a data lake) is designed to store large volumes of structured and unstructured data in-house.

Enhance the scalability by designing the data platform to follow a schema-on-read mechanism: data is ingested in real-time and only the required data is preprocessed prior to consumption for model training, analytics and designing.

Security and privacy by design

Modeling the physical design may involve sensitive personally identifiable information (PII). An example is the energy consumption in buildings — this can be used to accurately model the daily routines of the residents, which is certainly a privacy breach. Yet, this information is crucial for forecasting energy demand accurately for every building.

Considering any applicable privacy and security regulations, a cyber-physical system may incorporate mechanisms to mask user identity and anonymize all data before it is used to model and control any physical design attributes of the systems.

Architecture for cyber-physical systems

So how does a cyber-physical system operate with these characteristic features? That primarily depends on the application and the industry vertical.

Architectural frameworks for a cyber-physical system design commonly involve these components:

• Perception layer. The physical network of devices and sensors that produce raw information.
• Data transmission & management layer. The data is transmitted horizontally between physical devices and vertically across the Internet to a backend data platform for storage and preprocessing.
• Application layer. The software components that interact with raw information, train models, produce the digital-twins that can be controlled by an end-user.

For instance, the manufacturing and Industry 4.0 may follow a multi-level architectural framework. For example, the 5C Architecture containing five architectural levels (from lowest to highest):

• Connection. At this level, data is acquired from a smart network that consists of multiple devices and sensors.
• Conversion. Raw data is transformed into consumable information.
• Cybernetic. Network devices are virtualized and clustered for data mining and analytics.
• Cognition. Functions such as simulation and synthesis are integrated. A graphical user interface is developed for collaborative decision-making.
• Configuration. Autonomous self-configuration capabilities for resilience and robust system design are established at this level.

As we can see here, cyber physical systems can unlock a new path of innovation, with the internet as a messenger, what can we build in new ways?