Beyond Edge Analytics: Enabling Distributed Intelligence to Maximize real-time Collaboration across intelligent IoT network

0
327

By Prasenjit Bhadra, CEO, Ranial Systems
The exponential growth of interconnected intelligent machines, sensors and actuators has gained significant traction in transforming business operations. Such growing networks of physical objects are also facilitating the development of new breeds of products and services across various B2B and B2C domains. The proliferation of portable devices and high-speed wireless networks has been helping industries to develop scalable Internet of things (IOT) and machine-to-machine (M2M) infrastructure. With the wide-spread adoption of emerging IOT applications, organizations have realized that that the process of remote data collection and a centralized aggregation of the massive volume of such data or events wouldn’t deliver sustainable ROI in the long run. The urge of process innovation and autonomic functions across the subsystems demand terminal intelligence and real-time collaboration of the connected environment.

Current maturity of Cloud platforms, Big Data, and advance analytics have resolved the puzzles of streaming, analyzing and persisting massive volumes of sensory feeds, operational data, and events. Organizations have achieved near real-time visibilities of the automation environment and, are able to gain actionable insights. However, the aggregation of such raw feeds across multiple data islands and processing those to derive predictive and preventive measures result in an exponential rise in TCO (total cost of ownerships) of the high-performance cloud infrastructure and vulnerability of a centralized system managing most of the command, control measurement, and monitoring functions.

The emerging trend of edge analytics is ideally addressing some of these fundamental challenges by smoothening raw feeds, performing basic analysis functions and executing predefined actions. The evolution of powerful edge computing platform with a decent amount of storage and computing power has equipped IOT platforms to host sizable data, execute statistical or machine learning algorithms. The fundamental design principle of edge analytics should provision incrementally scale the collaborative functions across discrete automation silos, enhance operational intelligence, interoperability, and self-healing within the distributed IOT networks.

Where the next generation IOT systems are heading towards? Many mission and time critical applications demand situational awareness to interpret and act on certain physical conditions e. The latency of the network and growing complexities of centralized management functions fail to deliver a reliable control mechanism and on-demand interactions across multiple nodes within the specific zone of influence. Thus the intuitive design of the IoT gateway or edge controller has become an important consideration towards deploying a highly scalable and extensible IOT solution architecture. However, such ‘divide and rule’ policies should synchronize the run-time topology of the logical group of connected devices and management/ control modules deployed in the cloud environment.

Any scalable IoT infrastructure should leverage the distributed edge controllers or gateways as connected nodes within the mesh of synchronized grid logical functions that can scale and align with the disruptive operating environment. Thus the role of an edge computing device is not confined with the ‘store-n- forward’ data reliably and apply structured intelligence specific to data or instructions exchanged with the connected objects. The scope of edge analytics extends beyond defined set of smart operations. Each controller deployed at various levels of the hierarchy should be an integral part of the overall ecosystem and the flow of logical functions performed over time. The purpose-based design of hardware and gateway software silos increases complexities of integration of data and services across multiple functional units. Some of the most talked about features such as interoperability, self-healing, P2P collaboration would require adherence to a common standard of hardware, communication and software design specifications. The current state of implementation and middleware designs are treating each layer of the IOT Cloud in an isolated fashion while an integrated runtime topology of the technology and service deployments are becoming critical for deeper collaboration and enabling autonomic stability across sites.

The fundamental design constructs of any highly scalable IoT platform should consider an optimal level of abstractions access the hardware and software layers, which essentially help in provisioning implementation agnostic interactions over one or more communication channels and deploying need-based analytical functions in a standardized fashion. Such foundation could extend the possibilities of seamless integration of the segment architectures across the IOT application domains. Viz.; interfacing Advance distribution grid automation or connected health infrastructure within smart city ecosystem. A mature edge analytics platform should leverage context-aware data analysis and relevant command-control functions which can be synchronized with operating conditions and deployment semantics of connected physical objects. Such localized intelligence not only serves the real-time operations but also reduces the latency, cost of communication and network bandwidth requirements. Standardization of hardware and software platforms offers an extreme flexibility towards horizontal and vertical integration of data and services across the value chain. The power of edge analytics can be translated into an interconnected distributed intelligence wherein each node within the system can exchange knowledge and data with other nodes with minimal or no dependency of the centralized management hub. Such collaboration can expand the boundary of the autonomic and time sensitive operations across different sites; introduces an ideal level of redundancy to avoid the single point of failures.

Introducing a scalable foundation of distributed intelligence over the edge analytics platform optimizes the workload across the IOT network hierarchy. Defining an integrated view of software and hardware platform for any edge computing domain offers a well-defined separation of concerns across intelligent nodes from within various functional layers while standardization of core runtime ideally helps each node to replicate the whole or part of the operational behaviors of its peer nodes on demand. The supervisory functions of the centralized management hub would be provisioned to adapt meaningful information from the remote edge devices and deliver critical insights to the stakeholders in a timely fashion.

In a nutshell, the power of edge analytics could help us in building scalable operational layers which establish synergetic relationships across the data, knowledge, and functions to introduce intuitive workflows and management functions. A set of common services across the layers would help the nodes to learn and respond to the behavioral changes across the physical environment and, collectively exchange knowledge and data to take more informed decisions. The design principle of edge analytics foundation incorporating such critical functional or operational imperatives can meet the scalability and extensibility standards to maximize ROI over time and aid incremental innovation.