Originally printed in full in Object Management Group’s Journal of Innovation Feb., 2025: The Role of Data Centricity in Smart, Connected Systems – by Real-Time Innovations (RTI)
By Andy Krassowski, Senior Field Application Engineer, Real-Time Innovations (RTI)
Driven by AI and machine learning, the next wave of smart applications will rely heavily on fast and reliable data exchange across interconnected systems. Above all, this data needs to be secure, enabling different systems to communicate seamlessly. However, the complexity and scale of these modern systems can pose significant data connectivity challenges.
For safety-critical applications used in autonomous vehicles and surgical robotics – where every fraction of a second counts – ensuring low latency and high throughput for real-time AI processing can be a formidable task. Other challenges include seamless communication across diverse networks, scalability as systems grow, interoperability between hardware and software from different vendors, security risks during data transmission, and limitations in computing resources.
To tackle these challenges, a shift in system architecture is gaining traction: designing systems around data flow rather than processes. This approach is known as data centricity. By putting data at the core, systems can better meet the demands of smart, interconnected applications.
The idea of data centricity isn’t entirely new. Think back to the CRM systems of the 1980s, when customer data was centralized to create a “single source of truth.” Now, data centricity has evolved to operate at a much more sophisticated level, focusing on capturing data in motion for real-time communication in modern AI systems.
In a data-centric system, developers first identify the necessary data and then organize that data into Topics. The “how” and “when” of data delivery are managed by standardized Quality-of-Service (QoS) parameters, simplifying the initial design. The “who” (Publishers and Subscribers) and “where” (transport mechanisms such as Ethernet, Wi-Fi, or the internet) become more abstract. This data-centric approach offers important design benefits, such as the independent development of Publishers and Subscribers and transport-agnostic compute resources, all of which allow for greater scalability and flexibility.
Just over 20 years ago, the Data Distribution Service (DDSTM) standard was released. Expanded over the years, this proven, open OMGⓇ middleware standard effectively acts as the backbone for today’s data-centric architectures. DDS decouples data from the application, enabling real-time, scalable data exchange crucial for complex, high-performing systems across industries, including automotive, aerospace, healthcare, and industrial IoT. Unlike traditional client-server models, DDS provides direct access to structured data, fostering real-time decision-making and enabling point-to-point connections, thus improving scalability.
Key features of DDS include being transport-agnostic and, crucially, QoS policies. QoS allows developers to fine-tune system behavior based on application needs, configuring data reliability, fault tolerance, and performance guarantees to ensure critical data delivery on time and without loss. The data-centric communication of DDS simplifies development, improves maintainability, and enhances real-time decision-making. Furthermore, DDS supports compliance with other industry standards, promotes interoperability, helps avoid vendor lock-in, and reduces integration costs.
The functionality of DDS is enabled by a Discovery process in which communicating nodes exchange their Publisher/Subscriber details at startup. They then use these details to communicate efficiently via a binary structure using the Real-Time Publish Subscribe (RTPS) protocol. This protocol’s payload includes the data (Sample) and associated metadata.
Data centricity with DDS provides advantages over simple messaging systems. Through its QoS properties, DDS can understand the specific needs of multiple Subscribers and deliver only the relevant filtered data to each, saving both bandwidth and CPU resources. Another key feature is the handling of extended datatypes. When a datatype is updated, Subscribers can still process the base information and ignore new, unknown fields without crashing, providing vital future-forward flexibility for evolving systems. As new transports emerge, DDS can easily take advantage of them, as it has with the most recent DDS Specification, DDS-TSN.
Consider these examples:
- Automotive: DDS enables seamless, real-time communication between sensors (LiDAR, cameras, radar) and control units in connected and autonomous vehicles, ensuring the rapid flow of critical data for safe navigation. QoS prioritizes crucial data like collision alerts.
- Surgical Robotics: DDS provides the communication backbone for precise coordination between robotic arms, control units, and imaging systems. Real-time data from sensors is immediately transmitted, and QoS prioritizes critical data like force feedback and imaging updates, while also simplifying the integration of components from different vendors.
- UAVs: In defense systems, DDS facilitates reliable and secure communication across networks of autonomous vehicles and ground systems. Real-time sensor data sharing enables collaborative decision-making, and QoS prioritizes mission-critical information. DDS also offers platform-agnostic design and robust security features.
In summary, complex AI-enabled systems can benefit significantly from a data-centric architecture that is supported by the DDS middleware standard. Data centricity provides a firm foundation for building scalable, flexible, and more secure smart, interconnected systems to meet the needs of tomorrow.