By Edel Griffith, Technical Marketing Manager in the ASIC & IP Division of Adesto
We’re all familiar with the stories of exponential growth in the world of the Internet of Things (IoT). Millions upon millions of devices are expected to be attached to the cloud within the next few years. Basic intelligence can be added to any device quickly and easily, enabling it to become a member of the IoT ‘family’. What’s much less straightforward is trying to work out other immensely important aspects such as how to ‘retrofit’ security, determine communications protocols or deal with the way in which data is passed or handled.
These are the kind of problems likely to be encountered when presented with a technical scenario that was never foreseen and for which there are, essentially, no standards. One might say that it’s the biggest ‘bodge’ of all time. But the IoT also provides a massive range of opportunities. Because there are no constraints in terms of standards, developing IoT products allows designers free rein to interpret exactly as they want.
And of course that interpretation will vary enormously from designer to designer and company to company. Each will have its own way of looking at the problem it is trying to solve or service it is trying to provide. Different features will assume entirely different proportions and levels of importance for varying service providers, yet somehow all of these very different products and services need to attach to the internet via a server or a cloud-based service and be readily understood at both ends and interpreted correctly for the end user. A bewildering prospect indeed. The translation between these so-called endpoints can take place in the cloud, on the server or on the platform.
It may well be tempting to opt for using a ready-developed platform. Some are already in existence from the huge players like Google, Amazon and Microsoft and are termed platforms as a service (PaaS). These save customers from having to develop their own infrastructure and allow them to run and manage their services and applications relatively easily and without the hassle of having to maintain their own platform.
These cloud-based services are generally charged for on a per-usage basis. There can be problems associated with being ‘locked-in’ to a particular service provider or users may experience problems with service availability or network resilience but PaaS definitely provides a service that is much needed by many users at this relatively early stage in IoT development. It provides operating systems and much of the middleware users need to help them on their way. However users have also to consider what will happen if the service provider decides to cease support for a particular operating system or the programming language on which their service offering is based.
Because developers and companies will present their offerings so differently, the range of IoT solutions available will be vast. As the numbers grow, it is to be expected that the way they connect to the cloud will develop, with intelligence gradually shifting from centre stage, which it occupies at present, to the edge of the cloud.
At the moment, little processing is required or performed at endpoints. What processing there is tends to use legacy systems, bringing a new lease of life to them. Cloud-based systems are built under entirely different constraints. Performance is the key element for most users of the cloud.
As the migration of intelligence to the edge of the cloud occurs, so these two different pictures will merge, bringing a requirement for a mixture of the extant solutions. Developers of IoT edge systems will need to be able to create a server system that is capable of handling virtually any type of endpoint as well as creating and accommodating endpoints that are supremely flexible and adaptable. At the same time, users’ demands in terms of power constraints and availability need to be matched precisely with their requirements for performance.
Service providers of the future will need to be able to offer highly open architectures that can be extended virtually infinitely and that allow any type of device to connect to the cloud, no matter what protocol or communications technology it uses. At the same time, custom application-specific integrated circuits (ASICs) will be required to deliver whatever level of functionality the customer requires.
IoT endpoints will need memory chips that are capable of processing high levels of data on a local level. Energy-efficiency and high levels of performance will need to be harnessed on chips that also offer low levels of power consumption and low overall system cost. The expression ‘how long is a piece of string?’ comes to mind…
Companies looking for a partner to take them into the future of IoT could do worse than take a peek at Adesto Technologies, which has made a series of strategic alliances that will enable it to develop semiconductor and system solutions for IoT long into the future. The company is able to satisfy the needs of business users as well as those of the more demanding industrial marketplace. They may just be able to evaluate accurately the length of string that is required!