With system-integration being the most valuable and crucial phase in the development of virtually any IoT-based solution, it is imperative for IoT companies, and most importantly for those supplying IoT edge devices and infrastructure components, to constantly evaluate their product’s architectural options with respect to how they facilitate or, inversely, complicate inter-operability across the complete solution.
Development and adoption of IoT standards will play a fundamental role in enabling inter-operability, We believe that the approach taken by the oneM2M alliance, specifically focused on interoperability at IoT data level, goes in the right direction.
Beyond the obvious benefit of adopting relevant standards whenever possible in the IoT space, the inter-operability of components, sub-systems and services will be influenced by the following fundamental choices at product-architecture level :
The networking technology used to inter-connect the end-nodes among them and with their routers or gateways;
The choice of SW platform, including the embedded real-time OS, Middleware and SW development tools for feature-rich/smart end-nodes;
The data formats used to feed the applications running on the users’
There is indeed a very ample choice today for IoT edge networking technologies, and we believe that there is no obvious payback in investing R&D resources to develop proprietary solutions, or even to try to optimize existing standards, unless at least an order of magnitude improvement is obtainable in any of the key performance metrics, or substantial savings are obtainable in components’ cost.
Depending on the application, choosing the right networking technology can be challenging, and it is definitely one of the most critical architectural choices that system-integrators need to make. In certain situations, for example for electrically-plugged devices in Industrial and Building Automation, the use of power-line modems might be the simplest and the cheapest networking infrastructure fitting the bill.
The use of plain WiFi might be OK too if there are no stringent power restrictions, and even though WiFi might be oversized with respect to the data rates required to connect the end-points to the LAN gateway, it might be convenient for other reasons, like security, or simply because a WiFi network infrastructure is already in place or it can be used to co-network other, high-throughput devices for different applications.
In other situations, the end-points cannot be connected to the electrical distribution nor to a wired data network, and the edge networking choices are inevitably restricted to low-power wireless technologies. In these types of scenarios, and especially when the number of end-points to connect is high, one of the emerging options to consider is represented by the use of IPv6 over Low power Wireless Personal Area Networks, also referred to as 6LowPan for brevity. And in particular 6LoWPAN over IEEE 802.15.4 phy/mac layers, because this combination offers the best trade-offs in terms of scalability, energy efficiency, and performance, with the inter-operability advantage of a all-IP end-to-end solution. But there are many use-cases for large-scale IoT systems that may require the combined use of different networking technologies.
Companies that are positioned more down-stream in the value-chain, that is in the direction of the data coming from the IoT edge to the end-customers, have to consider how they facilitate the end-users’s access to, and consumption of data coming from the edge. Also at this level, for the Cloud Servers Infrastructure for instance, there are today many commercially available choices, generally known as Platform As A Service, and therefore developing a custom or proprietary solution in this segment might be justified only to support highly differentiated services, for example in order to control and guarantee very high levels of data security.
Since IoT is a huge and very diversified territory, even if many core technologies exist and are readily useable today, there are still several areas where innovation could drive significant value, beyond the system-integration space. It would be virtually impossible and definitely beyond the scope of this paper to try being exhaustive in listing all of the subjects where innovation would be beneficial. But here is a list with our current top-picks:
– Energy Efficiency
Energy efficiency is an area where there is virtually an endless need for innovation, and it affects every single system component of a IoT product or service, however computing and networking is where better energy efficiency brings the major pay-offs. Reducing the energy consumption of computing allows, in particular, to shift more intelligence towards the edge, which in turn reduces the need for end-points to transmit and receive application and control data from nodes that are more upstream.
Wirelessly-connected end-points, like sensing modules, clearly need to be as power-efficient as possible to extend battery life. More in general, multiple sleep/wake modes in all nodes and more intelligence in managing those modes can contribute very significantly to the reduction of the total power budget of the solution, because each node would be more energy efficient if it could go from, say, a deep-sleep mode to transmission modes only when there is useful data to be transmitted. But also because the energy required for the communication between nodes could be significantly reduced by optimizing the time each node spends just for listening for the transmissions from the other networking nodes it is connected to.
Battery technology itself and energy harvesting are other very valuable areas of innovation, of course: extending battery life is great, and being able to completely forget about batteries would be even better.
– Data-to-decision process intelligence.
Efficiency in data analytics, extracting value as quickly as possible from the IoT data is very valuable because time is money and because data patterns may evolve rapidly. More efficiency in pattern-recognition algorithms and processing, either in hardware implementations for increased energy efficiency and speed, or software implementations for increased flexibility would clearly help making the decision process at application level faster and more accurate.
– Security and privacy.
This is crucial area of concern and it is in fact widely considered as a major roadblock on the road to large-scale adoption of IoT technology and solutions. To illustrate how sensitive this point is, it is interesting to consider a prediction made by IDC in a study of December 2014, stating that within two years 90% of all IT networks will have an IoT-based security breach.
The good news is that in terms of information and computing technologies, again, much exists today, no radically new technical progress is required. With IoT being such a diverse territory, it is clear that there is no single universal solution to IoT security. The Trusted Computing Group has recently published a guidance document, which can be found here, for securing IoT. It is currently under public review, but it can serve as a good inventory of solutions available for a wide variety of IoT use cases. The challenge lies, once again, at system-integration level, where the overall coherent/trustable use of security techniques needs to cover the entire computing and networking chain from the edge, through to the internet and data servers, and to the users.
Legislation also needs to evolve and tackle specifically the IoT security challenges. There are in fact many relevant grey areas, or entirely missing ones, affecting privacy and security (the FTC November 2014 staff report is a very rich and useful reading on this subject). Which implies that, at least for the next 3 to 5 years, there will be virtually no limit to potential abuse in collecting and exploiting sensible data, nor valid methods or strategies for preventing and neutralizing security attacks or actions jeopardizing citizens’ privacy or the integrity of their personal information.
– Role of Open Source
Open source components are not only a way to accelerate innovation and share development costs, they also foster inter-operability across the board. The penetration of Open Source Software in the IoT space is already very wide and deep. If we consider for example the internet infrastructure alone, it would be hard to find today nodes that are not running some version of Embedded Linux and Carrier Grade Linux. Same goes for the Cloud servers and services, which are almost completely built with open source components and tools.
OpenSource hardware is emerging also in a number applications, and connected prototyping platforms can be very easily accessed. It is therefore of paramount importance for anyone with the plan and the ambition of building a new IoT product or service to systematically look for applicable Open Source components immediately at the start of prototyping, and adopt the best known practices to leverage and contribute to the Open Source community to make sure that critical R&D resources are invested only where new and differentiating components are valuable and functional to the target business.
IoT is a huge territory for developing and deploying advanced technologies, and it promises to drive booming growth in many markets, some of which will undergo fundamental changes, while other will appear because of the widespread use of internet and connected products and services.
Several challenges exist, from security to interoperability and legislation, but the potential benefits to address them are so high that a lot of resources will be dedicated to make the IoT a reality over the next 5 to 10 years. Most if not all of the building blocks to built IoT products and services exist, are easily accessible, and it is important to focus R&D and financial resources especially on increasing system-integration efficiency and know how.
Marco Carilli www.oriturtech.com Partner IAG e Massimo Vanzi Senior Member IAG