Concern about the Internet of Things (IoT) is largely focused on two key factors: a) the colossal number and b) the diversity of its new “users”. In population terms, the predictions range from 20 to 40 billion devices by 2020, and the questions include whether the Internet can handle so much traffic.
Even more serious is the diversity: to what extent can the same infrastructure accommodate users as different as smart meters, traffic monitors, medical devices, autonomous vehicles and high speed financial transactions… to name but a few?
The challenges, and the questions resulting, are daunting. So the first step has been to divide this influx into two broad categories. At the top end there will be many highly critical connected devices – such as heart monitors, industrial control systems and autonomous vehicles – each with their own specific demands for reliability and performance. Their basic question will be: “is the IoT communications good enough for us?”, says Brion Feinberg, vice president and CTO, Analytics Business Unit, Spirent.
It is a good question, because at the lower end there will be a far greater number of very humble devices – such as traffic monitors and smart domestic appliances – that won’t require the same level of communication service quality. It is estimated that these low-end devices will become the overwhelming majority, representing maybe nine tenths of the market – hence today’s emphasis on the massive market potential at the lower end.
However, even if the higher-end IoT only amounted to 10% of the $13 trillion total predicted by 2019, it would still present a massive market opportunity. Mobile service providers are well placed to exploit this opportunity because wireless carriers not only have the advantage of owning their own spectrum, they also have extensive experience managing large networks to deliver high reliability. To address this lucrative high-end IoT market, they will need to differentiate themselves from “best effort” services by offering superior network performance and security.
There will be wider benefits from this approach: because it often makes sense to address complexity by first developing solutions for the smaller population of smarter devices and then trickling down that knowledge and experience to develop simpler, cheaper solutions for the lower end devices.
What basic steps will be necessary to ensure reliable, high performance IoT communications that can serve and satisfy the SLAs for even the most critical applications?
To get this in perspective, go back to 2006 and tell the carriers that within ten years they would be expected to support across the globe over two billion highly sophisticated, data hungry, mixed media mobile devices called “smartphones”. They might well protest that this would be asking too much, and yet it has happened, and the service the carriers deliver is good enough to keep the market buoyant.
This miracle has been achieved by advances in technology combined with sophisticated monitoring of functionality, performance and quality of experience. When planning and developing any connectivity service, engineers will model the network in a laboratory and monitor it under a range of operating conditions – both normal and extreme – to ensure that it is reliable, secure and that it functions properly. In the case of a mobile network, this means not only emulating good operating conditions but also extremes such as heavy network usage, bad weather conditions – even cyber attacks on the network.
Nor would any smartphone manufacturer dare launch a new handset without first ensuring its operation in conditions that include poor network coverage, travelling at speed, multitasking, and combinations of these operating stresses. Meanwhile the network is not only monitored for functionality, but also for performance and other more complex QoE (Quality of Experience) criteria. We not only have sophisticated equipment for running these tests, but also have years of experience in anticipating possible problems and “knowing where to look”.
The network monitors’ nightmare
One of the most basic needs is to look for abnormal behaviour that might indicate a fault in the system, or a malfunctioning connected device, or even a malware invasion. To do that you first need to know what would be normal behaviour.
The smartphone is a highly complex device and its behaviour patterns in the hands of different users can be very different. But we now have years of experience of recognising what is or is not normal smartphone behaviour. If a particular smartphone is suffering, or causing, connectivity problems, then the monitor can access the device’s IMEI (International Mobile Equipment Identity) code and, knowing the make and model of the device and its characteristics, can readily judge whether it is working properly.
But when today’s IoT vendor needs a connectivity solution, it is common practice to buy an off-the-shelf IoT connectivity module. In this case, the IMEI data will announce that it is, for example, a particular Telit module that is on the network, but this does not indicate what sort of device it is connecting. It could be a smoke alarm, a parking meter, a heart rate monitor or some other unknown type of device – so what sort of normal behaviour should we expect?
A need to know
There is an urgent need for mechanisms to recognise IoT devices and monitor their communication – how often the devices are connecting, how many bytes they are sending to how many different destinations, etc. Ideally we need devices that provide the basic communication measurements to determine the network KPIs.Telcos would do well to insist on self-reporting capabilities in the devices., or at least to offer incentives such as lower certification costs to encourage their adoption.
It is vital that the IoT industry recognises the importance of ensuring that their devices’ actual communication performance data should be made accessible in this way. We should also be accelerating the recognition of industry standard categories of devices and their typical operating parameters, so that automated monitoring systems can detect anomalous behaviour – such as a refrigerator that begins transmitting spam mail.
When it comes to designing, implementing and integrating a large population of diverse systems to build something as complex as a “smart city”, even the simplest devices could trigger unanticipated problems. What would happen after a power outage, for example, unless there is some “random back-off” activation to reduce the risk of overloading the network while thousands of devices simultaneously reconnect?
Existing mobile network protocols require such a backoff, but implementations aren’t always perfect. And without some detailed understanding of all the elements being connected, their communication needs and expected traffic patterns, monitoring the network for abnormal or unusual traffic would be a nightmare.
What is needed is not just to provide connectivity for devices but also to make sure their identity and behaviour is not hidden from the network’s management and monitoring systems.
Work is being done to develop suitable standards. The Open Mobile Alliance has defined a Lightweight Machine to Machine protocol (OMA LWM2M) for managing IoT devices. As the name suggests this compact protocol is suited for simple devices with limited processing resources. This protocol already includes optional objects that the device can report that identify the device function and provide essential traffic measurements.
There is plenty of concern about the dangers of an IoT connecting millions of simple relatively vulnerable devices at the same time as ensuring very reliable connectivity for a smaller population of highly critical devices.
The good news is that there are solutions and methodologies already being used to ensure the highest standards of connectivity while automatically monitoring and managing networks as complex and demanding as those serving thousands of smartphones.
The technology is available, but it cannot work unless the network speaks to it. For example, we already have developer tools available to manufacturers that include functionality, across multiple communication protocols, for the device to provide identification and traffic measurements , enabling better communication between hosted devices and the network management.
Remember that the strongest long-term IoT value proposition is not about adding on silos of similar devices, but about the enormous potential for sharing data from so many diverse sources. The industry does need innovative techniques to make sense of so much diverse data, and it would be a good starting point to insist that no device can be connected that will be invisible to the system’s management and monitoring functions.
Existing mobile network operators, with their long history and deep experience delivering high quality communication services, should take the lead by insisting on devices that can provide device identification and traffic measurement capabilities for monitoring IoT device communications and assuring superior network service.
The author of this blog is Brion Feinberg, vice president and CTO, Analytics Business Unit, Spirent
About the Author:
Brion leads efforts across Spirent Communications to apply advanced data analytics for telecommunication service providers. In his role as vice president of product development, Brion has been responsible for the planning, implementation and delivery of the company’s products that provide customer experience management for mobile network operators. His focus has recently expanded to include identifying opportunities to apply analytics for managing the Internet of Things and monetiSation of customer experience network data
Comment on this article below or via Twitter: @IoTNow_ OR @jcIoTnow