Is there an Internet of Things?
In the late 1980s, the term
‘ubiquitous computing’ was coined by Mark Weiser, chief scientist at Xerox
PARC. He advocated that the best computer is a quiet, invisible servant. In
his vision, he described several scenarios that are still reference
scenarios for what is generally called the Internet of Things (IoT). How
much closer are we today to Mark Weiser’s vision, and to what extend is IoT
a new technology or just an evolution?
Recently, I read some old documents
from the late 1990s, originating in Eurescom projects, studies and workshops
that discussed smart devices and their integration into IT infrastructures.
It was interesting to read about the challenges and opportunities that the
authors were outlining in their reports and presentations. Many of them were
just extrapolations into the micro-world of appliances and were dealt with
largely as a problem for software. Today’s work on IoT still addresses
pretty much the same fundamental questions.
Large number of objects
The large number of objects continues to be a major concern. Related
deployment and management issues are being intensively researched. However,
if we compare the number of mobile phone subscriptions worldwide that,
according to the ITU, will reach 5 billion by the end of 2010, we have a
mismatch in perception. Furthermore, the number of mobile broadband
subscriptions will reach 1 billion at the same time. Such subscriptions are
used with so-called smart phones that include a number of additional devices
ranging from multiple wireless interfaces such as Bluetooth, WiFi or Zigbee,
to other sensor devices like GPS, accelerometers, gyroscopes, and others. If
I add the software capabilities of each smart phone, then the aggregation of
devices and software objects easily reaches the mark of 10 billion. So, it
looks like mobile network operators and device and equipment vendors have
today the technology and the business models in place to manage and operate
this very large amount of devices and objects.
The question that needs to be
addressed is the limit of growth. Today, all business models assume a
certain average revenue per user (ARPU). The sum of all revenues will reach
a boundary that is related to a certain percentage of the GDP of societies.
Whatever this limit is, it will be reached, and the only way to increase the
number of devices and objects will be to significantly lower the costs per
unit. These costs relate to both the capital expenses (CAPEX) for
developing, manufacturing and deploying the devices and the supporting
infrastructure as well as the operational expenses (OPEX) for operating
them. Hence, it makes sense to try to find ways to automate a large number
of steps required to manage the services. This is why the whole area of
autonomic computing and networking plays a vital role in the further
evolution of the IoT. However it should be noted that autonomics had emerged
as a topic due to the system complexities in most other areas of IT and is
not specific to IoT.
Security and privacy
Another area that needs significant research is the area of security and
privacy. Today, the security and privacy properties of almost all devices
and related services are left to the user who has to properly configure tens
or even hundreds of options manually that in most cases he doesn’t even
understand. The preconfigured options are mostly unsuitable. But is this a
new challenge? The ever increasing amount of software on the desktop and
services on the web have reached the limits of what a human is capable to
manage and configure. IoT makes a bad situation just a bit worse.
Energy and environmental
The requirement related to energy consumption of devices that need to
operate autonomously over longer periods of time is as old as the invention
of batteries and the first battery powered transistor radios. Unfortunately,
the technological progress in battery lifetime was very modest in comparison
to the evolution of micro-electronics, which significantly increased
performance. However, this had the effect that new developments were aiming
at increased performance for the same amount of energy, rather than trying
to increase the period of autonomous operation. This trend has changed
recently in particular due to the requirement to deploy devices that must
autonomously work for years or decades.
Additional requirements have
emerged, namely that some of the deployments of sensors are not recoverable
if, for example, deployed in the sea, desert or woods. These requirements
not only call for very energy-efficient micro- and nano-electronics, but
also for bio-degradable materials, including electronics and batteries. The
contribution of IoT with respect to this topic will mostly come from
material research and less from software optimisation.
Trying to analyse the different arguments of why the Internet of Things is
something different, I failed to find something that justifies the creation
of a new technology domain. IoT is for me indeed an extrapolation of today’s
and tomorrow’s technologies into ever smaller artefacts. If we keep treating
IoT as something special, we risk developing new technology fragments that
we would have quite a hard time to integrate back into the infrastructure of
the Future Internet as it evolves. So, I suggest to stop treating IoT as a
new technology domain. The Internet of Things is just a marketing term and
should be treated as such.
send us your comments
on this article.