IoT devices are all around us, and will communicate 400 zettabytes of data (that’s 4GB with 22 zeros after) by the end of 2018, but an AI component is needed to make sense of all this data - otherwise we end up with nothing more than a mountain of ones and zeros.
The increase in processing power brought by AI, combined with this data, will finally allow connected technology to reach its full potential. This is especially important given the shrinking size of modules, which will allow devices to proliferate and in turn feed more data back to AI. By including Artificial Intelligence within networks, numerous problems facing businesses today can be solved, and IoT and AI can grow stronger together, for our benefit.
Smart vs. Simple
Telecoms are already using AI to maintain infrastructure and secure traffic (more on this later), but what about connectivity? The problem with connectivity modules is that they can only support the most basic classification functions of Machine Learning. AI programs therefore run on network servers, allowing them to analyse metadata (device behaviour) from millions of sources. MVNOs (Mobile Virtual network Operators) are better positioned to use AI in this way, to manage complicated network infrastructures and select the best data path.
Multi-network SIMs have been around for years, but more advanced switching methods need more advanced computing. Multi-IMSI (International Mobile Subscriber Identity) connectivity allows the SIM to switch to an entirely different network infrastructure, so data is not routed back through a ‘single point of failure’ at the top.
Multi-IMSI technology uses one SIM profile, so that means it can also be programmed onto one ‘slot’ of the eUICC standard. eUICC SIMs have distinct SIM profiles and can fall back to a ‘bootstrap’ profile that provisions additional networks over the air (OTA). This gives a resilient ‘failover’ profile that is not tied to one operator - especially important considering the bootstrap must be provisioned at the point of manufacture for embedded eUICC SIMs, regardless of where the device will be deployed.
Maximum resilience is ideal for mission-critical, remote, and roaming applications, but others require even more AI alchemy to switch between technologies. Sigfox’s hybrid chipset, announced late 2017, connects their technology, NB-IoT, and LTE-M. This means a device can transmit basic data on the Sigfox ultra-narrow band, and switch to a cellular network to send more information when necessary. Allocating these data packets to the right network is perfect for AI and could show us what this technology can do for the IoT.
You scratch my back…
The development of new technologies, and integration of networks with each other to provide an open data space, opens up an intriguing possibility for AI to grow. As neatly described by AT&T’s Mazin Gilbert: ‘AI feeds on data just like a car feeds on gas’, and the telecoms industry can happily provide the data AI needs. What makes the IoT such an interesting training ground for AI is that the two main datasets - device interactions and identification - are not linked to individuals, meaning that working with this data does not affect end-users.
By combining these datasets, AI is able to predict when devices need maintenance (by analysing an identified device’s behaviour), can detect possible security breaches, and respond with appropriate action before anybody notices. IoT security is already changing with AI, and old procedures (detect bug, inform company, wait, announce the find) are being reevaluated. One ‘next gen telco’ feeds this stream of new bug announcements into an AI algorithm, to help preemptively develop patches for new bugs that emerge.
Together, for better
As enterprises adopt new network technologies for different parts of the business, these systems will have to communicate effectively, and not create new problems for IoT applications. AI will help untangle these systems, especially when applied to a network layer that can provide deep data penetration for AI to sink its teeth into and will also benefit from being fed such a huge amount of behavioural data.
Applying AI to big data systems like this is not widespread, the telecoms industry could get far more from AI than just customer service chatbots, even cutting away the complexity to find easier, more direct ways to communicate between machines. As IoT systems get even more complex, to satisfy an ever-growing range of applications, AI will automate processes that will soon be too complicated for even experts to handle - and in turn this data will help AI grow, and be applied to all number of industries.
About the author: Charles Towers-Clark is the founder of Pod Group - a platform and communications company established in 1999 providing data connectivity and billing for the Internet of Things. Through his work, Charles has become expert in IoT, machine learning and artificial intelligence. Charles is now writing a book on the impact of AI in the future world of work.
Edited by
Ken Briodagh
