The IIoT opportunity seen from a signal-processing perspective

Address challenges involving real-time signals exchanged by many data-generating devices.

By Susanna Spinsante, PhD May 9, 2017

In the future, process-production and discrete-manufacturing industries increasingly will rely on the Industrial Internet of Things (IIoT) to improve their operations. While there are many definitions for IIoT, one of the more common is "a distributed network of smart sensors that enables precise control and monitoring of complex processes over arbitrary distances."

A 2016 survey by PriceWaterhouseCoopers (PwC) revealed that 33% of leading industrial and manufacturing companies with current high levels of digitization are projected to increase their digitization efforts to 72% by 2020. The survey of more than 2,000 participants from 26 countries showed that leading investment areas include vertical value-chain integration (72%), product development and engineering (71%) and customer access, including within sales channels and for marketing efforts (68%). Almost 72% of manufacturing enterprises surveyed predicted their use of data analytics would improve customer relationships, and 35% of companies adopting the European variant of IIoT, Industrie 4.0, expected revenue gains higher than 20% over the next five years. These results illustrate the growing expectation that IIoT will be the driver for third industrial-innovation wave.

But what does IIoT mean from a signal-processing perspective? 

Signal processing’s role

The IIoT concept includes all the main characterizing features of the IoT computing paradigm, including ubiquitous sensing, data interaction and collection and data analysis. These functions are enabled by machines that talk to each other as they complete tasks—in a smarter and more efficient way than possible by humans acting alone. Machine-to-machine communication supports autonomous communications among devices. It enables collaborative automation between machines and intelligent optimization of industrial processes. Autonomous machine capabilities enabled by IIoT culminate in cyber-physical production systems (CPPS). In other words, systems for which the boundary between what is physical and what is digital becomes increasingly indistinguishable.

Signal processing plays a critical role as an IIoT enabler, despite being overshadowed by other aspects of IIoT, including the communication architectures, sensing technologies and power management involved. Advanced machine-learning approaches will support predictive and prescriptive analytic solutions by connecting previously stranded data from smart sensors, equipment and other assets. This eases the means to enable condition monitoring, failure diagnostics, efficiency improvement and downtime reduction. By anticipating failures, these approaches assist with continuous improvement at both design and manufacturing levels.

The IEEE Signal Processing Society (SPS) IoT Special Interest Group (SIG) promotes the development, standardization and application of signal- and information-processing technologies targeting unique challenges from emerging IIoT scenarios. These challenges, among others, include: analyzing, summarizing and protecting real-time signals and information exchanged by massive data-generating devices—including sensors, machines and robots—and their corresponding data-processing nodes. 

Focus on the industrial

Significant industrial sectors impacted by IIoT include process production, discrete manufacturing, utilities and oil & gas. However, these "industrial" applications also have relevance in areas such as "smart cities" or "smart agriculture" domains. The common foundation of these sectors relies on the basic integration between information technology (IT) and operation technology (OT), enabled by the IIoT.

The IIoT market is estimated to reach near $124 billion by 2021. According to Ovum, the more remarkable investments today are taking place in industrial settings such as manufacturing, transportation and utilities. Automotive and consumer IoT, as well as ad hoc IoT-enabled smart-city projects are emerging. Typical cross-industry IoT use cases include smart lighting and smart traffic solutions in cities, intelligent machine applications, condition monitoring, smart agriculture, and healthcare. These cases are on the rise and expected to grow even more in 2017 under the IIoT umbrella.

As IT and OT converge, IIoT technology enables more innovative, on-demand approaches to customer outreach. As a result, new—and sometimes unexpected—revenue and business models emerge.

Yet despite innovation and increasing awareness about IIoT opportunities, the industrial sector lags in fully embracing IoT for reasons as follows:

  • Because long-lifecycle legacy field devices won’t be upgraded anytime soon, to achieve ubiquitous IIoT the solutions must be "appendages" to in-service products. 
  • The scaling of vertical IIoT solutions presents challenges.
  • Engineering managers resist deploying unproven technologies. 

Energy consumption

Even In its early stages, IIoT has been used to reduce resource consumption and carbon emissions by industrial systems. Nevertheless, IIoT systems—including a diversity of devices with sensing, processing, and communications capabilities—consume substantial amounts of energy, which can contribute to a larger carbon footprint. On the other hand, IIoT systems typically consist of low-power devices supported by batteries, which constrain the continuous operations of IIoT systems.

In the IIoT domain, data collection relies heavily on massive numbers of sensor nodes and smart devices. Thus, optimized IIoT sensing, processing and communications may effectively reduce energy consumption. Acting as the backbone of IIoT systems, wireless sensor networks are the main source of energy consumption. Therefore, signal-processing techniques—aimed at more efficient radio transmission, communication protocols and use of the shared radio resource—become critical and have a practical impact. Similarly, the design of efficient power management algorithms for systems and nodes supplied by batteries is essential to ensuring a long lifetime for the industrial infrastructure.

Even multimedia signal processing contributes to IIoT development, despite appearing to be a technology quite far from this domain. Multimedia signal processing leverages new technologies, including gaming, augmented and virtual reality, 3-D displays and wearable devices for new workers who must be trained or those who are already employed within the production plant. High-fidelity simulation-based applications for training, or for real operation on the plants and their assets, require extremely efficient algorithms to process multimedia signals.

They must perform at high speeds due to the real-time operational constraints involved; with high precision in terms of synthetically localizing the user into a virtual scenario reproducing the plant; and with high responsiveness, due to the very short delay between the action performed by the user in the virtual domain and its effect on the real, physical plant. Running these immersive simulations improves learning and aids in developing skills to deal with unanticipated plant situations. Workers increase their confidence in performing the job functions assigned to them and dealing with emergencies. Other simulation applications include testing and validating new software and supporting system migration.

Raw and real-time data from sensors and end nodes—and aggregated data from information sources like intermediate systems and devices, which can be queried via built-in capabilities—fuel the development of robots capable of taking specific actions. IIoT becomes a driver of autonomous decision-making devices. Such an Internet of "robotic things," already in place in big warehouses, requires development of extreme machine-learning algorithms to support intelligent robots that can self-assign tasks and operations and make decisions.

The usage of the IIoT, within a broader context, will ultimately lead to connected ecosystems that encompasses supply chains, factories, etc., thus giving the extended enterprise concept a new meaning. 

Signal-processing perspectives

Unique signal-processing needs arise from the IoT ecosystem. These include: robust information sensing from complex and adverse environments using massive connected sensors and distributed signal processing; low-power situation-aware data-transmission and processing; and the privacy-preserving processing of information that is shared by connected things.

In IIoT systems, data transmission and its sharing among machines is critical to the performance of the whole system. Therefore, research and investigations in this area are gaining interest as new machine-to-machine standards and protocols emerge, combined with affordable sensing and communication modules. However, due to the complex system structure and heterogeneity of hardware and software platforms, ubiquitous access and interoperable sharing of the machine-generated data are still open issues. An underlying standard messaging mechanism-based on cross-platform technologies to support the communication between machines-is a significant concern to the success of autonomous industrial systems to ensure proper quality of the collected data and information.

In April 2016, Morgan Stanley published the results of a survey from which data security and cybersecurity emerged as growing concerns for organizations relying on universal connectivity, as is the case for industrial applications enabled by IIoT. These environments typically feature hybrid and mixed connectivity solutions that range from cellular and low-power wide-area networks to industrial connectivity solutions that require innovative approaches to data and communication security, far beyond the traditional firewall-based solutions used in legacy network infrastructures. In this context, signal processing helps companies design new security tools that are IIoT-compatible, based on the distributed-ledger concept at the foundation of blockchain, or exploiting efficient and robust mathematical primitives, such as elliptic curves.

Signal-processing aspects of IIoT will see increasing adoption in 2017 and beyond. These new manifestations might include mobile robots able to collaborate with humans to perform complex tasks; wearable computing platforms for industrial environments; additive manufacturing; and IIoT technologies that drive increased supply-chain visibility. In their turn, IIoT solutions will support a set of flourishing areas, like asset-monitoring and tracking, smart grids, digital oil fields and smart buildings, with an initial focus on energy management. 

Edge device evolution

Edge devices or intelligent gateways also play a prominent role in the growing IIoT infrastructure and these network devices will be used to collect, aggregate, filter and relay data close to industrial processes or production assets. By running analytics and advanced machine learning algorithms, they detect anomalies in real time communicate to operators. Therefore, the emerging trend is to move intelligence to the edge of the network, closer to the data sources. This is also a viable option when it is not feasible to run analytics on a cloud platform or when a cloud-based solution is unavailable.

In IIoT, the edge of the network is being populated by devices, including in embedded systems, and heterogeneous communication technologies that range from Ethernet connections to wireless and cellular gateways. Protocol conversion gateways can interface disparate networks in the framework of the emerging device-to-cloud integration trend. Fog computing and streaming data analytics will leverage IIoT to feed any sort of artificial-intelligence application.

IIoT is going to dominate the evolution of manufacturing and industry in the coming years. It will be driven by the increasingly vanishing boundary between the physical and the digital worlds. New approaches will tackle the many challenges and opportunities: from data curation and processing, to advanced and efficient communication technologies, to extreme machine-learning algorithms. It is important now for the engineering community to foster research and innovation in this exciting field by supporting actions that highlight the importance of signal processing, in its many manifold flavors, for improved productivity in industrial environments.

The role of the IEEE Signal Processing Society

IEEE’s Signal Processing Society (SPS) is the world’s premier association for signal-processing engineers and industry professionals. Engineers around the world look to the Society for information on the latest developments in the signal-processing field.

  • The society’s deeply rooted history spans almost 70 years, featuring a membership base of more than 19,000 involved signal-processing engineers, academics, industry professionals and students.
  • The purpose of SPS’ IoT Special Interest Group is to promote the development, standardization and application of signal- and information-processing technologies targeting unique challenges from emerging IoT scenarios that require analyzing, summarizing, and protecting of real-time signals and information exchanged or shared by massive data-generating devices such as sensors, machines, robots, cars, etc., and their corresponding data-processing nodes.
  • Examples of unique signal-processing challenges from IoT include: robust information sensing from complex and adverse environments using massive connected sensors and distributed signal processing, low-power situation-aware data-transmission and processing, and privacy-preserved processing of information shared by connected things. 

Susanna Spinsante, PhD, is senior member of the IEEE and member of Signal Processing Society’s Special Interest Group on IoT. She’s also an RTDA (temporary researcher) at the Università Politecnica delle Marche, Ancona in Italy. You can reach Susanna at s.spinsante@univpm.it.

Callouts

Even multimedia signal processing contributes to IIoT development, despite appearing to be a technology quite far from this domain.

The design of efficient power management algorithms for systems and nodes supplied by batteries is essential.

Machine-to-machine communication supports autonomous communications among devices.

This article appears in the IIoT for Engineers supplement for Control Engineering 
and Plant Engineering

– See other articles from the supplement below.