The Internet of Things, which may deliver the type of “auto-magical” level of control over home and business environments promised by science fiction authors for decades, is held together through wireless connections. It is a source of data roughly proportional in volume to the area it serves and its importance to the human operators in and around it.
It follows that IoT data is a marketable commodity. People who use connected devices to run their business need to know who’s interested in their IoT datasets, and whether this interest may be detrimental to their interests.
The Spy Cleaning Your Living Room
Those who own or plan to own a Roomba, a smart floor-cleaner with connectivity to other devices that improves with each iteration, may have been alarmed to learn that their robot vacuum can generate data-based maps of their living spaces, and that Roomba’s manufacturer, iRobot, plans to sell that data to big tech companies.
This is just one basic example of how connected-device data may be mined and monetized. The devices in a typical IoT field can generate a great deal of data in a range of types and formats, including status, time-stamps, metadata, sensor data covering parameters such as temperature, pressure and air quality and logs.
This list will continue to grow as IoT devices evolve in terms of function and type of connection. IoT data may be valuable in itself by depicting the functions and efficacy of ecosystems such as supply chains, but it also has value in a context of research and development – existing or ongoing sets of IoT data can contribute to improvements in the next generation of connected devices and systems, as well as helping develop new solutions for IoT data storage, security or analysis.
However, IoT data also may be a target for hackers and others with interest in data theft.
For the moment, it appears that the economic value that can be derived from IoT data sales could be vastly more exploited. According to a recent estimate, the data from barely one in 100 sensors in the IoT field of oil rigs is used to generate value for the industry. Equally, models based on IoT data analysis could prove valuable to companies that have not yet implemented IoT technology.
Third-party data could contribute to the development of other models used to predict failure in systems, maintenance needs and sensor placement in order to cover a given area most efficiently. This means data may be of interest or importance to a broad range of companies, infrastructure authorities or other government bodies.
The Coming IoT Data Marketplace
Therefore, the scope for an IoT data marketplace is clear, particularly at a time when big data is growing so rapidly. Such a platform may be one on which providers amass and collate huge amounts of data, turning it into unified finished products for sale – ideally not including individual or corporate identifiers.
IoT data management may also be sorted into various categories, such as sensor data, based on what the devices in question track or record. The value of this marketplace could be of the same order as those for IoT devices – the sensor market is projected to reach $1.4 trillion by 2020.
There is a demonstrable demand for IoT data aggregation in order to construct products for this marketplace. IoT devices do not typically store much data locally; information is usually sent to a cloud or a local database, depending on how quickly it needs to be retrieved. The data can be collated using tools such as Hadoop or NoSQL databases. Alternatively, companies such as ThingSpeak offer products known as “channels” through which IoT data can be sent.
There are various basic types of IoT data aggregation types: centralized, cluster-based or branching. Data aggregation is also conducted within networks to provide adequate and timely syncing between devices. In addition, some researchers argue that it also optimizes the volume of transmissions between individual devices, and thus enhance the longevity of individual IoT networks.
Localized aggregation may also reduce the power consumption needs of IoT, making it another potential value point in the IoT data ecosystem.
Analysis and Interpretation
The value of the best aggregated data in the world may be limited without further processing to enable proper interpretation. In the case of IoT, analysis often must be conducted in real time due to the nature of its connections and functions – knowing what to do next – but it can also create archives that require high-throughput analysis as the volume increases. As a result, IoT data analysis has become more extensive in its approach compared with other, traditional information forms, employing methods including distributed analytics, edge analytics, machine learning and real-time analytics.
Distributed analytics are essentially optimized to match the topography of the IoT field in question, and can also be supplied by Hadoop tools such as Apache, but need to be matched by distributed aggregation. Edge analytics are applied to Internet of Things frameworks such as those of Linux at the interface between an internal IoT and the wider internet, and can conserve bandwidth between the two forms of network.
Machine learning is a particularly well-suited form of analytics for both real-time and archive-stored IoT data collection, capable of being scaled up if necessary and of developing its own parameters in response to factors such as the increased volume of connections or devices over time. Machine learning may also be more versatile in terms of the models that can be applied to the data and can identify its own variables if necessary.
Real-time analytics – often time-sensitive and requiring reduced latency – are also highly suited to IoT data, and also can be supplied by Apache frameworks such as Samza or Storm.
The framework and type of analytics for any IoT data type needs to be carefully matched to the use case to ensure that the data is processed adequately, and to make it more useful to a potential purchaser.
The Data Ownership Conundrum
All this means that retaining ownership (or in some cases privacy) of IoT data management is likely to become an increasingly pressing concern. It could be ensured by implementing blockchain technology to connect and govern an IoT field, authenticate data and eliminate the risk of theft or erasure. It could also interact with blockchains of other companies or datasets within collaborations or partnerships.
Alternatively, these functions could also be governed by artificial intelligence applications that could, when needed, activate security protocols, authorise IoT data sale or use by third parties, and interact with other AI systems automatically.
The Internet of Things may account for as much as 10% of all the world’s registered data by 2020. It may directly or indirectly depict a company’s systems, fittings, assets or infrastructure, having been designed to monitor and maintain these attributes. IoT data of similar types may be aggregated to form sets for research and development, and may require analysis conducted with respect to time frames or other data types.
All this may represent the basis of a market for IoT data where the customers are likely to include groups such as Google, Facebook and Amazon, as well as academics and governments. Therefore, companies may need to determine or secure the ownership of their IoT data analytics and storage as they do with any other asset, through technology such as blockchain. In this emerging environment, IoT data analysis, collation, processing and monetization may represent significant new opportunities for companies that understand the nature, application and value of the data they hold and generate.
