Special Feature
Part of a ZDNet Special Feature: Sensor'd Enterprise: IoT, ML, and big data

Enterprise IoT projects: Data, ML, security, and other key factors

The Internet of Things has been hyped, discussed and piloted for years, but is now beginning to deliver real business benefits.

Enterprise IoT projects: Key factors for successful deployments The Internet of Things has been hyped, discussed and piloted for years, but is now beginning to deliver real business benefits.

The collection and analysis of data from sensor-equipped devices in order to achieve a business or organisational goal -- a.k.a. the Internet of Things, or IoT -- is a key component in the wave of digital transformation underpinning the Fourth Industrial Revolution.

Although the IoT has been discussed and analysed for many years (ZDNet's first special report on the subject was in January 2013), there's a widespread sense that the pieces are now falling into place for it to begin delivering real value for businesses of all kinds, and not just early adopters. Business value will flow from real-time information about operations, supply chains and customers, which (if analysed properly) should translate into lower costs and increased revenues. Better information about business processes should also lead to lower environmental impact and wiser investment decisions.

A trillion IoT devices

arm-annual-iot-devices.png

Image: Arm

The scale of the upcoming IoT disruption was the subject of a white paper published in June 2017 by chip designer Arm, entitled The route to a trillion devices: The outlook for IoT investment to 2035. By 2035, according to Arm's analysis, the IoT's boost to global GDP will be $5 trillion, the annual spend on IoT hardware and services will be $1 trillion, the cumulative spend on IoT connectivity modules from 2017 will be $750 billion, and 1 trillion IoT devices will have been built since 2017.

Here's Arm's projected breakdown of that $1 trillion annual IoT spend in 2035:

arm-annual-iot-spend-20352.png

* Microcontrollers, apps processors, radio controllers, memory; ** Sensors, batteries, solar cells, antennae, circuit boards, etc.

Data: Arm / Chart: ZDNet

With $100 billion a year potentially on the table for digital electronic components alone in 2035, the reason for Arm's interest in the IoT is very clear. Firms involved in IT services ($450bn/yr), telecoms services -- including 5G and its successors -- ($150bn/yr) and installation services ($100bn/yr) also have a lot to play for.

Here's Arm's projected breakdown of the 1 trillion IoT devices that it predicts will be built between 2017 and 2035:

arm-trillion-iot-modules2.png

Data: Arm / Chart: ZDNet

Different IoT modules have different connectivity and power supply requirements, which will affect their unit cost: according to Arm, the cheapest (with a BOM of around 40 cents in 2017 and 15c in 2035) will be a smart tag powered by RF energy harvesting with NFC or RFID connectivity, for example, while the most expensive (BOM around $8 in 2017 and $3 in 2035) will be an IoT gateway powered by mains electricity with internet access via an unlicensed radio connection such as wi-fi or Bluetooth.

SEE: The rise of industrial IoT (ZDNet special report) | Download the report as a PDF (TechRepublic)

And where will that $5 trillion global GDP boost come from? Prominent sectors in Arm's breakdown are: food production and distribution (food waste reduction, water/fertiliser/pesticide reduction, yield increases); manufacturing (throughput increase, preventative maintenance, after-market revenues); wholesale and retail (targeted advertising, inventory management, supply-chain management); transport and logistics (fleet management, asset utilisation, fuel savings, paperwork elimination); healthcare and social assistance (preventative medicine, drug research, home care, patient monitoring); and government, education and defence (improvement in traffic monitoring, crime prevention, pollution control, waste management).

Fog and Edge: Data processing for the IoT

If this kind of IoT expansion comes to pass, the amount of data available for analysis will be enormous, and much of it will require processing in real time -- or at least with low latency. This has prompted the realisation that the traditional data-processing model -- sending all data to centrally located (on-premises or cloud) data centres -- will need retooling. Two contenders have emerged: 'fog' computing and 'edge' computing, both of which bring processing capabilities closer to the data sources, thereby reducing traffic to core data centres, decreasing latency and accelerating response times for critical applications.

Special feature

Special report: Sensor'd Enterprise: IoT, ML, and Big Data (free PDF)

This ebook, based on the latest ZDNet/TechRepublic special feature, explores ways IoT is improving operations and delivering business value to enterprises around the world.

Read More

Fog computing is a Cisco idea, backed by the OpenFog consortium (founding members Arm, Cisco, Dell, Intel, Microsoft and the Princeton University Edge Laboratory), whose mission statement reads (in part):

"Our efforts will define an architecture of distributed computing, network, storage, control and resources that will support intelligence at the edge of IoT, including autonomous and self-aware machines, things, devices, and smart objects. OpenFog members will also identify and develop new operational models. Ultimately, our work will help to enable and drive the next generation of IoT."

Edge computing is a similar idea, promoted by (among others) the EdgeX Foundry, an open-source project hosted by The Linux Foundation. EdgeX Foundry's goals include: building and promoting EdgeX as a common platform unifying IoT edge computing; certifying EdgeX components to ensure interoperability and compatibility; providing tools to quickly create EdgeX-based IoT edge solutions; and collaborating with relevant open-source projects, standards groups and industry alliances.

According to EdgeX Foundry, "The project's sweet spot is edge nodes such as embedded PCs, hubs, gateways, routers, and on-premises servers to address key interoperability challenges where 'south meets north, east, and west' in a distributed IoT fog architecture":

edgex-foundry-iot-schema.png

Image: EdgeX Foundry

EdgeX Foundry's technical steering committee includes representatives from IOTech, ADI, Mainflux, Dell, The Linux Foundation, Samsung Electronics, VMWare and Canonical.

IoT security and the rise of blockchain

The biggest worry with the rise of the IoT is security: what kinds of havoc could bad actors wreak with billions of internet-connected devices -- in homes, offices, factories and other critical infrastructure -- at their mercy? We don't have to use our imaginations, as serious IoT-based cyberattacks have already happened. Perhaps the most notorious was the 2016 attack that used Mirai malware to recruit hundreds of thousands of poorly-secured Linux-based IoT devices into a botnet, which then launched large-scale DDoS attacks on high-profile websites and service providers.

The likely growth in the number of IoT devices and the pressing need to secure them has led analyst firm Gartner to predict that spending on IoT security will total $1.5 billion in 2018 (up 28% from 2017) and reach $3.1 billion by 2021 (up 107% from 2018):

gartner-iot-security-spending.png

Data: Gartner / Chart: ZDNet

"Although IoT security is consistently referred to as a primary concern, most IoT security implementations have been planned, deployed and operated at the business-unit level, in cooperation with some IT departments to ensure the IT portions affected by the devices are sufficiently addressed," said Ruggero Contu, research director at Gartner in a statement on 21 March. "However, coordination via common architecture or a consistent security strategy is all but absent, and vendor product and service selection remains largely ad hoc, based upon the device provider's alliances with partners or the core system that the devices are enhancing or replacing."

A widely discussed solution to the problem of securing IoT devices, and the data they generate, is blockchain -- the distributed ledger system that's best known for underpinning cryptocurrencies like Bitcoin. While blockchain technology might not prevent the factory-default credentials of cheap IoT devices being accessed by botnet-creating malware, it can help to create networks of trusted devices that exchange data securely and trigger automated actions via 'smart contracts'.

That's the goal of the Trusted IoT Alliance (TIoTA), a consortium comprising blockchain technology companies, enterprises (including Bosch, Cisco and Gemalto) and IoT technology providers, which launched in September 2017. The TIoTA notes that "Thus far, the Internet of Things has been deployed without much trust in the provenance of device identities, integrity of device software, or verifiability of device data," and defines as its primary purpose "to leverage the blockchain and other security technologies to introduce trust into IoT, to leverage the automation potential of trusted IoT sensors with smart contracts, to evaluate the ROI of use cases enabled by trust, and to create common open source methods in the smart contracts as a system of lego blocks."

tiota-reference-architecture.png

Image: Trusted IoT Alliance

In February this year the TIoTA published its Trusted IoT Reference Architecture, following this up in March with the Trusted IOT Alliance Testnet, a global private IP network with member-managed endpoints where any protocol can run and connect to both the internet and the private network.

All this sounds promising. However, in its predictions for 2018, specialist IoT market research company IoT Analytics sounded a cautionary note: "At this point, we believe that it could easily take 5+ years before the [blockchain] technology gets used to secure individual end-to-end IoT applications in the field or before a significant number of devices autonomously engage in a smart contracts-based data exchange."

IoT and AI/ML

Exponential growth in the number of IoT devices will require new data-processing architectures and serious attention to security. But perhaps the key link in the value chain will be the application of artificial intelligence (AI) and machine learning (ML) algorithms to extract actionable insights from the resulting flood of IoT data. These algorithms could be deployed at the edge (flagging up and transmitting anomalous data patterns, for example), or at the core (analysing medium/long-term trends, for example).

IoT data will come in different volumes, varieties and velocities, and, as the authors of this February 2018 research paper note, the broad goal is to use data mining and AI/ML algorithms to uncover patterns and generate insights in the most efficient manner possible.

In their paper, Mahdavinejad et al provide a listing of the 14 most common supervised, unsupervised and reinforcement ML algorithms for classification, regression, clustering and feature extraction. These algorithms range from K-Nearest Neighbours and Naive Bayes to One Class Support Vector Machines and Feed Forward Neural Networks, via Linear Regression and Principal Components Analysis, and are matched to typical IoT and smart city use cases.

The main takeaway from this study is that different IoT applications involve different numbers of devices and types of data, which generate specific features that will be best characterised by applying the most appropriate ML algorithm -- turning 'big' data into 'smart' data in the process.

What does all this look like in practice? IoT platform provider C3IoT offers a good example in the nuts-and-bolts area of inventory optimisation, where the disparate data sources include "demand, supplier orders, production orders, bill of materials (time-varying), change history of re-order parameters, and inventory movement data". Using the company's AI-driven C3 Inventory Optimization application, a global manufacturer of complex equipment achieved a 30 percent reduction in inventory levels and projected annual savings of $100-$200m:

c3iot-inventory-example.png

Image: C3IoT

"Thanks to the intersection of today's elastic cloud, big data and IoT technologies, combined with the application of powerful AI methods, manufacturers can now finally realize the promise and benefits of dynamic, adaptive inventory optimization," the blog post concluded.

IoT deployment: the state of play

Cisco

cisco-iot-report-cover.png

At the Internet of Things World Forum (IoTWF) in May last year, Cisco presented the results of a survey covering 1,845 IT and business decision-makers in enterprise and mid-market companies from the US, UK and India. Six industry verticals were represented: retail/hospitality, energy, transportation, manufacturing, local government and healthcare. All respondents worked in organisations that had already completed or were developing IoT initiatives, and all were involved in the strategy and direction of at least one IoT initiative.

The resulting report, The Journey to IoT Value: Challenges, Breakthroughs and Best Practices delivered a surprising headline finding, given the amount of hype surrounding the IoT in recent years: only around a quarter (26%) of surveyed companies were successful with their IoT projects -- although as Cisco's IoT marketing chief Inbar Lasser-Raab noted in her IoTWF presentation, only 15 percent were "truly failing".

That leaves around 60 percent of companies who were neither succeeding nor 'truly failing' with their IoT projects, so it's no surprise to see statements such as these emerge from Cisco's survey:

  • 60% believe that IoT initiatives look good on paper, but prove more complex than expected
  • 64% agree that learning from stalled or failed initiatives help accelerate their IoT investments
  • 61% believe they have barely begun to scratch the surface of what IoT can do for their business

What did Cisco learn from the 26 percent of companies that did achieve IoT success? Key factors here were: good collaboration between IT and the business; a technology-focused culture; and IoT expertise gathered via internal and external partnerships.

Generally, IT executives were more positive, with 35 percent considering their IoT initiative a complete success, compared to 15 percent of business executives. There were differences in emphasis too: IT types stressed the importance of technologies, organisational culture, expertise and vendors; meanwhile, business execs were more interested in strategy, business cases, processes and milestones.

On the partnerships front, successful organisations were more involved with the partner ecosystem at every stage of their IoT projects:

cisco-iot-success-factors.png

Image: Cisco

If we look at the 'delta' between most successful and less successful organisations, it seems that partner involvement at the strategic planning stage of IoT projects is currently most important:

cisco-iot-partner-delta2.png

Data: Cisco / Analysis & chart: ZDNet

When it comes to 'blockers' holding up the progress of IoT projects, the key factors in Cisco's survey were time to completion, quality of data, internal expertise, IoT integration, and budget overruns.

Despite the relatively low overall percentage of successful projects in its survey, 73 percent of Cisco's respondents felt that IoT data was benefiting their businesses in areas such as: improved product quality or performance; improved decision-making; lowered operational costs; improved or new customer relations; and reduced maintenance or downtime.

Cradlepoint

cradelpoint-report-cover.png

With help from Spiceworks, Cradlepoint -- a provider of cloud-based WAN networking solutions for enterprises -- surveyed 400 IT professionals in the US, Canada and the UK for its State of IoT 2018 report. Respondents were all involved in some capacity with IoT strategy, and worked for companies with at least 500 employees. Industry sectors represented were: manufacturing, education, IT services, healthcare, government, retail/wholesale, financial services, construction, telecommunications and energy/power/utilities.

In Cradlepoint's survey, only a third (32%) of organisations said they currently used IoT, although more than two-thirds (69%) had adopted or planned to adopt IoT solutions within the next year. As with Cisco's survey summarised above, the broad-brush picture here is one of a technology area that's on the cusp of lift-off.

Supporting that interpretation is the fact that just 27 percent of respondents identified IoT as a top initiative for the coming year:

cradlepoint-it-initiatives.png

Data: Cradlepoint & Spiceworks / Chart: ZDNet

Given that IoT projects require considerable re-engineering of IT infrastructure and for security and analytics to be in place, it's no surprise to see these initiatives ahead of IoT in the IT priority queue. In fact, when evaluating IoT projects, the main factors considered by Cradlepoint's respondents were security (41%) and return on investment (35%).

Ironically, although security is a major worry with the IoT it's also a top driver for IoT adoption (along with improved operational processes, reduced OpEx, simplified management, reduced IT complexity and improved flexibility): in Cradlepoint's survey, 71 percent of respondents said they were using IoT for building security applications. Yet it was insecure IP cameras (and other IoT devices) on internet-facing networks that allowed attacks such as Mirai to occur, fuelling fears about the IoT. Even so, nearly half (49%) of Cradlepoint's respondents said their IoT systems would reside on core enterprise networks, rather than a separate network dedicated to IoT technology (40%).

Another concern raised by Cradlepoint is the extent to which companies plan to implement IoT systems in-house, rather than partnering with external providers. In all eight stages covered by the survey, in-house implementation trumped external partnership:

cradlepoint-iot-implementation2.png

Data: Cradlepoint & Spiceworks / Chart: ZDNet

Looking at the 'delta' between these numbers, IoT data storage emerges as the most in-house-biased stage, followed by data extraction, solution building and data analysis:

cradlepoint-iot-implementation-delta2.png

Data: Cradlepoint & Spiceworks / Analysis & chart: ZDNet

"Organizations that plan to implement, house, and manage IoT in-house are taking a back-to-the-future approach," concluded Cradlepoint, echoing the finding of Cisco's survey that IoT projects are more likely to succeed if companies involve the partner ecosystem rather than try to go it alone.

Cradlepoint concludes with some best practices that, it says, "will allow companies to mitigate the potential for a massive security incident and increase the odds of achieving ROI on IoT systems":

  1. Treat network security as a foundational consideration from the inception of the planning process, not as an afterthought.
  2. Do not try to implement IoT applications using only in-house resources and IT generalists. Work with one or more trusted partners/vendors with IoT expertise to drive initiatives forward effectively.
  3. Consider whether legacy network infrastructure -- which requires manual, error-prone, and time-intensive network segmentation and policy orchestration -- can really meet the needs of this fundamentally different technology.

Outlook

The Internet of Things has been hyped, discussed and piloted for years, but is now beginning to deliver real business benefits. However, with surveys revealing only moderate success rates for IoT projects, there's clearly work to do in deploying suitable edge computing architectures, setting up trusted (blockchain-mediated) data flows, and applying the most appropriate AI/ML algorithms to extract actionable insights. The IoT is a complex area and, it seems, businesses would do well to look beyond their in-house resources in order to maximise the chances of a successful result.

READ MORE ON THE INTERNET OF THINGS

What is the IoT? Everything you need to know about the Internet of Things right now
The Internet of Things explained: What the IoT is, and where it's going next.

What is 5G? Everything you need to know about the new wireless revolution
It is a capital improvement project the size of the entire planet, replacing one wireless architecture created this century with another one that aims to lower energy consumption and maintenance costs. Sure, you'll download movies faster on your phone, but that's not the real reason 5G's arrival is being accelerated.

IoT in the office: Everything you need to know about the Internet of Things in the workplace
Bringing the Internet of Things into your workplace can bring new opportunities -- and some risks too.

The success of your IoT initiative will depend on your data proficiency
Forrester highlights takeaways from Bosch Connected World and the importance of data proficiency in IoT initiatives.

Machine learning and the Internet of Things
Gartner predicts that more than 65 percent of enterprises will adopt IoT products by the year 2020.

Most IT pros fear IoT cyber attacks. Few are doing anything about it
A report from Ponemon Research and Shared Assessments suggests a major disconnect when it comes to IoT security.

Cybersecurity in an IoT and Mobile World
The technology world has spent so much of the past two decades focused on innovation that security has often been an afterthought. Learn how and why it is finally changing.https://www.zdnet.com/topic/harnessing-iot-in-the-enterprise/

Harnessing IoT in the Enterprise
The tech revolution is spreading to every corner of the earth with the Internet of Things, and it's enabling data analytics and automation in ways never before imagined in business.

How to use IoT to save money on your office bills (TechRepublic)
Companies can explore different options for smart lighting, HVAC, and security systems to cut down on operating costs.

NVIDIA and ARM want to bring deep learning to your IoT projects (TechRepublic)
Announced at GTC 2018, the partnership will see the NVIDIA Deep Learning Accelerator architecture integrated into Arm's Project Trillium.

How SMBs can maximize the benefits of IoT initiatives (Tech Pro Research)
IoT isn't just for the big players-in fact, smaller companies often have better agility to implement practical initiatives for both internal operations and commercialized solutions.