“Search engines have come a long way since their original purpose of locating documents, but they still lack the connection between social, physical and cyber data which will be needed in the IoT era,” said researcher Dr Payam Barnaghi. “IoT data retrieval will require efficient and scalable indexing and ranking mechanisms, and also integration between the services provided by smart devices and data discovery.”
“I see tremendous opportunities to effectively utilise physical data, especially IoT data, with cyber and social data by improving the abilities of machines to convert diverse data into meaningful abstractions that matter to human experiences and decision making,” said Professor Amit Sheth of Wright State University in Ohio. “IoT search, particularly for devices or machines to interact with each other to find and aggregate relevant information on a human’s behalf, will become a critical enabler.”
The IoT will see more machine-to-machine searches which are automatically generated depending on location, preferences and local information. “Autonomous vehicles, for example, will need to automatically collect data such as traffic and weather information from various sources without a user being involved,” said Barnaghi and Sheth in ‘On searching the internet of things: requirements and challenges’, an article published in IEEE Intelligent Systems.
Requirements will include being able to access numerical and sensory data, and providing secure ways of accessing data without exposing the devices to hackers, Said Surrey University: “Applications which rely on public data, such as smart city technologies, need to be very accessible to make them available to a wide range of applications and services. Search mechanisms for these devices will need to provide efficient methods of indexing, crawling and finding data while ensuring the data is safe from hackers.”
The University of Surrey’s 5G Innovation Centre has a number of IoT search engine projects – including developing search mechanisms that describe the sources of the data required, and developing algorithms for clustering and analysis of IoT ‘time-series’ data.