The Web, from People to Objects
[ C H R O N I C L E ]
First let us recall an important yet little-known fact. Tens of billions of microprocessors are produced per year, but most of them are integrated into objects of every sort: household appliances, audio-visual equipment, cars, etc. The chips in our computers or laptops only represent 2% of the total (although much more in value). Even smartphones, now widely available, “only” represent two billion chips per year. For a long time, object IT remained poor, with rudimentary programming and a slew of obscure formats for their data interfaces. Luckily, a profound change is now underway.
Designed in CERN at the end of the 1980s to enable better computer communication between humans, the Web is now also the standard communication platform for humans with objects, themselves increasingly connected, and for objects with each other. This major evolution is better termed the “Web of Things” than the more traditional “Internet of Things” (IoT); although the internet is the main communications network between computers, the Web (which we use as an internet application) offers much more, with a standard document publication interface (web pages), data exchange and programme downloading.
At the start, the Web was created for human clients calling servers which sent back simple display pages, without keeping transactions in memory. But with constantly increasingly needs, it underwent several evolutions: parameters transmitted in HTML transactions, memorisation of client data systematically sent back to the servers (cookies), addition of executable code in the pages, first simple (CGI), then random and able to access the client machine (JavaScript), encoding of data structures or random programmes (XML, JSON), etc. Now Web transactions do not even need human action; a client programme can automatically dialogue with several servers or with other clients, a server programme can interact in a coordinated way with other servers and clients, etc. Complex and geographically distributed activities can thus be controlled using the Web.
Although this programming uses a set of ad hoc and assorted technologies, it is far more systematic that with the proprietary interfaces of yesteryear. Furthermore, and essentially, to access our connected objects, we need nothing more than a web browser or a standard telephone application. The Internet however is far from an ideal platform. The speed and latency cannot be guaranteed, nor even the arrival of messages other than statistically, although they may be crucial for the applications. In another field, it is essential that a car’s brake pedal communicates with the brakes, securely and with a well-controlled message transmission speed; this requires a quality-of-service guarantee on networks, found in upmarket cars (TTP, FlexRay) and that will probably become widespread. As the Internet is open by nature, security is a considerable issue and still only poorly mastered. Attacks on things can be even more devastating than those on personal data. The massive attack on part of the Ukrainian electricity network at the end of 2015 is unfortunately probably only a foretaste.
Much remains to be done therefore, and the marvels boasted everywhere in the hyper-connected, reliable and secure world of the internet of things will not be easy to achieve. It is not yet excluded that security issues make life almost impossible for the web of things, especially for things that have low energy. If it succeeds however, the benefits could be considerable throughout society, science and technology, thanks to a wide and varied field of possible inventions that is only just beginning to emerge.
> AUTHOR
Gérard Berry
Computer Scientist
Gérard Berry is a professor at the Collège de France, member of the french Academy of Sciences and CNRS gold medal in 2014.
