IoT (Internet of Things) platforms bring together the tools to churn out digital services. The data extracted from connected devices and sensors will feed a torrent of new applications development aided by IoT platforms. Digital malls, venues for clusters of app stores where services of a kind, such as smart city applications, will sell services.
While IoT platforms aggregate the building blocks for producing digital services, software-defined networks will be their supply chains. Virtualized interfaces on software-defined networks will draw on a vast pool of microservices and other inputs, available on networks, to mediate the production of digital services.
The complexity of IoT platforms
In their mature phases, an array of IoT platforms will help to fuse together the resources and data from multiple industries for services generation. IoT platforms range from those which connect thousands of devices, to application enablement, analytics and machine learning, and vertical-specific application development.
IoT platforms are prodigious creators of applications as they interweave a diversity of resources and skills. Smart cities, for example, will pull together data from the connected vehicles industry, geo-location information services, cloud computing, and that from connected street lights for traffic management. “The hallmark of IoT is that it is a system of systems and complex ecosystems associated with them,” said Joseph A. di Paolantonio, an independent IoT analyst with DataArchon.
Digital services producers will benefit from data network effects, on top of the network effects of software-defined networks, as the volumes and variety of data extracted from a diversity of industries, joined together with the help of resources like APIs available on IoT Platforms, explode. The number of users and the data generated by them will multiply as an increasing number of digital services are generated.
After years of efforts to connect devices, sensors, and networks, IoT platforms are making early efforts to build complex systems for applications enablement. The Exhibit A for such complex systems is Nokia’s IoT platform for smart cities. It is a unified platform, for a gamut of smart city services, which integrates with Nokia’s Integrated Operations Center (IOC) for the orchestration of multiple smart city services. The data will be extracted region-wide from sensors placed in the towers of telecom service providers. For analytics, it has co-developed a suite of smart city analytics services with StarHub of Singapore. The payments for services available at the smart city digital services mall are secured and mediated by a blockchain.
Itron is an example of accelerated application development that happens when business domains intersect. For some decades, Itron’s business focused on the energy and water sectors. The turning point came when it was able to connect devices with a wide-band mesh network with a higher bandwidth of up to 2.4 Megabits per second while comparable networks offer 50 Kbps.
The mesh network became the bedrock for its expansion into smart cities. As LED lighting grew in cities, the mesh network helped to interconnect street lights and create a network that covers the entire city. “This created demand for a very large spectrum of new smart city IoT applications. In order to provide all these services, we opened our IoT platform to developers,” Itai Dadon, Director of Smart Cities and IoT at Itron, told us. “We now have an extremely rich ecosystem of partner applications and devices that we can sell via our platform,” Itai Dadon informed us.
An unexpected outcome of an open IoT platform was that an innovative startup,Databuoy, developed an application, aided by Itron’s SDK, for gunshot detection based on research completed at the Department of Defense. Unlike existing similar applications, it does not require human intervention to isolate gunshots from other explosions such as from firecrackers. “The microphones, small enough to be embedded in street lights’ luminaires, recognize gunshot signatures, and listen to their propagation from signals transmitted by multiple street lights to confirm they are gunshots, not other blasts all without human judgment,” Itai Dadon explained to us.
Itron’s repertoire of applications expanded with the extended reach of its IoT platform. It now includes traffic management, parking services, seismic activity detection, gas leak detection, and infrastructure monitoring.
Data gathered at the edge and from devices feeds the development of new applications. One of these services is the estimation of the time duration of urban journeys. “The traffic data collected at the edge is processed and displayed on digital twins, with virtual objects like street lights, and meshed with GPS data to estimate the time duration of car journeys after factoring in the time intervals of changes in street lights,” Itai Dadon revealed.
Learning from the data uncovers unforeseen opportunities. Foghorn’s Lightning IoT platform, for example, started with performance monitoring and predictive maintenance in industrial plants. Over time, it learned that the return-on-investment on the IoT platform is far higher when CEP (Complex Event-based Processing) based analytics is combined with machine learning to gain process efficiencies. “Machine learning in combination with CEP based analytics is used to find ways to reduce scrap, improve yields and to minimize machine downtime in factories,” Sastry Malladi, CTO, and co-founder of FogHorn, told us.
Connectivity and data extraction
The use of IoT platforms has accelerated as the initial challenges of connectivity of devices have been overcome. Mesh or peer-to-peer networks, widely used for connecting devices, encountered challenges with network traffic management in the absence of routers and centralized control. For overcoming them, “we have self-healing mesh networks which re-route traffic automatically in a dynamic setting such as an urban environment,” Itai Dadon of Itron told us.
The environment for embedded devices is markedly different and connects with the edge or wide-area networks after its adapted for the purpose. “Industrial systems use many different protocols other than IP and Ethernet which need connectors, to be interoperable, which our IoT platform provides,” John Petze, the co-founder of Skyfoundry told us.
The extraction of data from embedded devices has its own sets of challenges. “Most companies don’t have experience with machine data which has billions of pieces of time-stamped data processed in milliseconds. They do relational databases which are not suitable for machines,” John Petze informed us. “The design of analytics software is for small footprints of distributed devices like Raspberry Pis; most engineering graduates train on servers,” John Petze added.
Connectivity platforms strive for ways to interconnect across a variety of radio protocols and cellular connection options which have proliferated as LP-WAN networks have grown. Wirepas is one company which reduces the frictions in the flow of data when a variety of radio protocols are used. “Wirepas is radio-agnostic — it provides software to other platforms or connectivity solutions to create a mesh network regardless of the radio protocol in use,” Joseph A. di Paolantonio told us.
There are other companies specializing in cellular IoT connectivity and communications — Telit, Sierra Wireless and Gemalto.
Breakthrough connectivity will be achieved when LP-WAN networks connect across a region. Traffic congestion management, for example, will anticipate choke points with greater precision when data on vehicle flows is available for a downtown area but also inflows of traffic from out-of-town vehicles, sports stadiums, and adjacent neighborhoods.
Filament, a startup, uses higher frequency radio spectrum for short-range radios inter-connected with mesh networks and conversely lower frequency radio bands for long-range communication across mesh networks all with an ad-hoc networks of interconnected devices. To cope with the security risks of movement across local networks, it uses blockchains for secure passage.
Application enablement with IoT platforms (AEP) is the next stage, and there are early signs of progress on this count. “IHS started a specialized research and advisory service covering IoT platforms in 2018 in response to the rising demand for information and data in this market,” Sam Lucero, Senior Principal Analyst for IoT at IHS Markit told us. IHS estimates that revenue for AEPs will grow at the rate of 33.1% compound between 2017 and 2022 and rise to $83.0 billion in 2022
Customers are beginning to recognize the complexity of IoT applications development and the need for support from partners. A recent survey found that 57 percent of the decision-makers for IoT projects reported that they use IoT platforms while another 35 percent planned to do so. “At the outset, the bias is to develop applications in-house. IoT platform vendors tell me that their best prospects are those who have tried to develop IoT applications and realize how hard it is to scale before they reach out to a platform vendor,” Sam Lucero informed us. “There is still market education happening to grasp the specifics of the complexity of the technology and the details of the way to address it,” Sam Lucero surmised.
Some sectors like smart cities, automotive, manufacturing, and consumer have experienced relatively higher rates of adoption. “Larger volumes and longer commitments by customers have motivated vendors to offer attractive discounts,” said Sam Lucero. Concurrently, vertical platforms like Siemen's MindSphere have gained traction.
Data from the control plane which helps to determine service quality and performance has been more widely used compared to that from the data plane. The business case for the latter is more challenging as the data is more varied.
Another class of companies, several of them among the cloud services providers, have gained traction with their horizontal IoT platforms which provide the common denominators in vertical applications. These include device connectivity solutions, security, and remote monitoring of devices.
The most complex are the diagonal IoT platforms that are a hybrid of the vertical and the horizontal and are likely to be most productive in application development as they will process data from both types of sources. Amazon, in combination with Alexa for Business and Greengrass for edge computing besides the AWS cloud, and Hitachi Ventara, which merged Hitachi’s data management, IoT, Big Data analytics, and edge computing businesses to provide operations technology and IT solutions across multiple verticals, are examples of companies offering diagonal IoT platforms. Amazon has gained momentum with its voice solutions and Hitachi’s IoT analytics reported robust growth of 134 percent in bookings for 2016.
“Diagonal IoT platforms have been difficult to scale, primarily because of the need to monetize, manage, and establish digital trust across the intersections of such complex ecosystems,” Craig Bachmann, advisor to TM Forum, told us. The TM Forum has conducted workshops on monetization of IoT platforms, especially focusing on connected ecosystems addressing “diagonal” business scenarios. The richest set of IoT opportunities result when “industry verticals” cross over in a “diagonal pattern.” Among many possible examples, one is a Smart Car connecting to Insurance to Smart Health and to a variety of smart “diagonal” applications. These workshops have revealed the need for a common language, a common understanding of value streams, and a set of templates that comprise a new type of Ecosystem Business Architecture.
Complex ecosystems revolve around diagonal IoT platforms and are difficult to manage unless they are either led by a dominating partner or have established both a clear business operating environment and new diagonal IOT platform technologies. New sets of business capabilities have been imagined that enable the scaling of these diagonal market opportunities--digital service malls.
Distributed ledgers are among the capabilities under exploration. “Blockchains are a promising tissue that could hold together ecosystem partners and smart contracts, written in software, could enable scaling across ecosystems. There is still quite a bit of debate about smart contracts in the legal communities, but it is likely that computational law will become a norm in IoT,” Craig Bachmann concluded.
Analytics in the IoT environment
Diagonal IoT platforms will also face formidable challenges with analytics in an environment where data is sourced from hundreds of sources. Smart cities, for example, will gather data on air quality from hundreds of sensors spread over roads, factories and commercial buildings, and homes. They will like to know how the air quality is affected by emissions from vehicles, and efficiencies in energy usage by households, commercial offices, and factories.
The data will be collated from thousands of sensors and devices using a myriad of protocols and technologies. It will be categorized for analytics purposes; the circumstances in which the data was gathered would have to be considered before it is comparable across sources. Finally, it will need to be aggregated into higher level categories, such as air quality in downtown areas versus in suburbs, to gain insights.
“Today it is hard to prepare time-series of data let alone correlate it with location data to gain insights,” Joseph A. di Paolantonio informed us. Without the historical and location data, it is hard to answer simple questions such as changes in air quality over time or the extent of variations in air quality across downtown locations compared to the commercial districts.
“No one, at this point, is doing meta data and master data in a satisfactory way,” Joseph A. di Paolantonio observed. As a result, it is hard to be confident about data quality. For example, the sensors, or their technology and protocol, used for gathering data maybe swapped and their changed feeds could be misconstrued as an error at the aggregation point. Metadata, which accounts for the context of when, where, whey, how, etc., categorizes data independent of the technology used for sourcing it.
Similarly, master data management categorizes data for analytical purposes. For example, it will group data to measure emissions by vehicles for time-of-day or location. That will yield knowledge on the impact of road congestion on air quality.
Insights that lend themselves to actions, in real-time, need to be collected at various levels. “Most data today is aggregated and processed at the cloud level. It should also be collected and processed on devices and at the edge,” Joseph A. di Paolantonio. This will help, for example, in traffic management and answer questions about the choke points in city lanes and their impact on air quality or their impact on heavier vehicles compared to passenger cars.
The digital services malls of the future will grow as the interdependent elements of technology architecture, analytical design, ecosystem management, and business value creation are aligned with tools available on IoT platforms. In fits and starts, the future of IoT platforms is taking shape.