Source : Mouser Electronics

Because the IoT is fundamental to our daily lives, engineers are working hard to implement systems that ensure it keeps operating even when things go wrong.

Industry, healthcare, and energy utilities are using the Internet of Things (IoT) to transform the world. For example, industrial automation relies on the IoT to enhance productivity and product quality, while healthcare leverages it to dramatically improve patient outcomes while reducing costs. Moreover, utility companies are building smart grids that use IoT to multiply the reliability and flexibility of energy supply while heading for “net zero” electricity generation.

However, this heavy reliance on IoT poses certain risks. Three key threats exist: disruption to critical systems by malicious actors; unwelcomed users accessing the network to steal valuable personal data; and failure by complex technology supporting IoT. Imagine, for example, the chaos resulting from sections of electricity distribution systems experiencing unintended shutdowns in winter, the upheaval generated by customers’ banking details being compromised, or the mayhem caused by a central control failure of a hospital’s drug infusion pumps.

Engineers have worked hard to mitigate the security issues associated with hackers intent on disruption or data theft. Technologies like encryption and authentication—provided it is deployed properly and well maintained—can keep the door closed on all but the most determined bad actors. But such is the world’s increasing reliance on IoT that we need to do more than lock people out. The answer is to build-in system resilience so that if things do break we are not faced with catastrophic failure.

Taking an Alternative Route

At its most fundamental level, IoT is a massive network of interconnected “things”—devices with sensors, software, and many more communications systems. IoT relies on each thing on the network being able to share data, information, and instructions with every other thing.

In the early days of IoT, engineers attempted to lower latency, reduce energy consumption, and boost the throughput of the network by implementing systems to select the shortest single route between any two devices attempting to speak to each other. However, for things on different sides of the planet, for example, even the optimum route might still comprise thousands of wired and wireless links and nodes, every one of which would need to operate flawlessly for the data packets to get through.

If just a single link in the data exchange route failed, the connection between the two remote devices would be severed. That’s an inconvenience if one thing is a lowly sensor on the edge of the network, but if the device is a router, controller, or server at the center, then the impact becomes a lot more serious.

Today, greater resilience underpins the network by ensuring multiple routes are simultaneously mapped out between two devices in communication. Hence, if a link in the original path breaks, packets are quickly switched to an alternative route. The downside to this multipath approach is an increase in latency, greater energy cost, and a reduction in maximum throughput, but the tradeoff is worth it for a more robust network.

Building in Redundancy

Adding to the multipath approach, resilience has also been built into IoT architecture by duplicating critical components, resources, or data to mitigate the impact of failures. For example, redundant sensors, actuators, or controllers sit alongside their active partners ready to take over should failure occur. The same is true of software, where copies of datasets, applications, and services are maintained and activated when the original versions are corrupted or otherwise compromised.

A less costly approach to system resilience through redundancy can be achieved by constructing a back-up system of more modest capability than a complete duplication of the entire network. While offering a reduced service compared to the primary system, the back-up system could nonetheless be used to provide continuity for critical services in the event of main system failure.

Disaster Recovery as a Service (DRaaS) takes a different approach to redundancy by offering a commercial cloud-based service that provides automated backup, replication, and redundancy capabilities for key parts of IoT, particularly those parts used by industrial and commercial organizations. DRaaS helps organizations recover from network disruptions without high levels of expenditure on their own redundancy systems.

Predictive Maintenance for IoT

In an industrial IoT implementation, sensors continuously monitor a machine or process while an edge computer analyses parameters such as vibration, temperature, or electricity consumption for irregular trends. As a result, predictive maintenance is a key application for IoT. The detection of irregular trends triggers preventative measures to eliminate instances of catastrophic breakdown.

Now, predictive maintenance techniques are being applied to the network itself to enhance resilience. One technique uses a trusted connected server to collect and analyze application layer data from numerous sensors, actuators, controllers, and edge computers. This information is then used to train machine learning (ML) models that describe the typical behavior of the individual IoT devices. Once the ML models are refined, it becomes a simple task for the trusted central device to detect anomalous device activity and advise repair before data corruption or network failure.

Similar systems can also be used to mitigate security threats. Intrusion detection and prevention systems (IDPS) software monitors network traffic, system undertakings, and user behavior to look for signs of activities that seem to deviate from normal behavior. If something unusual is spotted, the IDPS steps into action, blocking malicious traffic, quarantining infected devices, and alerting supervisors.

Conclusion

Industry, healthcare, and commerce’s reliance on IoT is on such a massive scale that simply attempting to defend against outside attacks is no longer sufficient. The hardware and software complexities of IoT mean that communication failures, equipment breakdowns, and data corruption are inevitable. That in turn means that mechanisms are needed to prevent catastrophic failure when things go wrong.

Engineers are working hard to implement architectural enhancements to the network to make it far more resilient to breakdowns. Multipath networking, redundancy, system degradation, and DRaaS are helping to ensure that problems are isolated when they occur, thus preventing widespread communication breakdowns that might otherwise undermine productivity, dent economic activity, or even endanger lives. Furthermore, by embracing the power of ML, network technicians are now implementing predictive maintenance techniques to ensure that breakdowns don’t occur in the first place.