The Role of Data Centricity in Smart , Connected Systems
This article discusses how data centricity , enabled by core connectivity standards , enables realtime communications for advanced AI and autonomous systems .
Data-centricity is the underlying mechanism for developing a data-centric architecture . This approach emphasizes the central role of data in designing , implementing , and managing distributed systems . While there are a few different architectural approaches that use data centricity , this paper will focus on the Object Management Group ® ( OMG ® ) Data Distribution Service™ ( DDS ) protocol . DDS is widely used in complex AI and Machine Learning ( ML ) use cases , providing the communication layer that keeps data as the focal point , thereby enhancing system flexibility , scalability , reliability , data interoperability , and digital connectivity for next-generation systems .
Readers will learn how DDS decouples data from the application , which enables real-time , scalable data exchange for complex , high-performance systems . This paper concludes with an overview of how data centricity works in three industry use cases .
1 CONNECTIVITY CHALLENGES IN SMART , CONNECTED SYSTEMS
Next-generation AI and ML applications are beginning to deliver transformative value through intelligent decision-making , automation , and optimization . These advanced applications require rapid , reliable data exchange across a complex interconnected system of devices and subsystems to achieve stringent performance and safety requirements . Examples include surgical robotics , software-defined vehicles , autonomous systems , and military defense systems , where unreliable data flow could have severe consequences . In addition to low latency , the data must be protected against unauthorized access and be able to exchange information across systems , regardless of platform . At their core , smart applications require access to data - lots of data - as well as reliable connectivity to process that data rapidly , robustly , and at scale .
The complexity and scale of this effort introduces data connectivity challenges for developers of AI-enabled applications . To start with , the data in AI / ML subsystems can reside in various locations , including edge devices , centralized cloud platforms , and / or in central compute locations . Data exchange between these subsystems can occur internally within a system , across different operational technology ( OT ) systems , or between IT and OT systems ( e . g ., analytics platforms ). While all smart applications face challenges such as data integration , latency , and security , there are additional interoperability requirements when bridging IT and OT systems , due to differing transport protocols , hardware architectures and programming languages . A datacentric architecture can provide the framework to ensure cohesion which bridges these divides .
For applications where data delays pose safety risks , such as self-driving cars , it ’ s critical to ensure low-latency , high-throughput data exchange to meet demand for processing data in real time . Modern systems require seamless communication between multiple nodes , which can become problematic in environments with unreliable or high-latency networks . Scalability is another concern , as the addition of devices , sensors , or applications often introduces bottlenecks or
Journal of Innovation 3