Instrumentation & Measurement Magazine 25-4 - 22

one to achieve high computing capabilities by using several
low-cost processing units. Therefore, the capability which
distributed systems have to share resources involves the reduction
of the overall cost of the system. Moreover, distributed
systems are intrinsically more robust and reliable than concentrated
ones: generally, a distributed measurement system is
able to work properly even though one or more nodes fail. This
is something which, of course, cannot be obtained in a concentrated
system.
Distributed measurement systems are also scalable and, in
general, easily re-configurable. As an example, it is possible
to add or remove a measurement node without changing the
design of the whole system. In addition, nodes can be moved
from their location and deployed in other points without impairing
the function of the remaining part of the system. Such
characteristics and features are extremely difficult, or sometimes
impossible, to achieve with a concentrated measurement
system.
Unfortunately, all of these advantages come with drawbacks
and technological limitations. Some of such limitations
are related to the network employed for connecting the measurement
nodes. Specifically, the network infrastructure,
either if already existing or created ad-hoc, has to be available
in all locations where the nodes are deployed. Such a network
should be reliable as much as possible, and often nodes should
be able to operate even in case of network failure to guarantee
a minimum level of service and function according to the application.
Another important aspect related to the network is
its throughput which is always limited by physical and technological
factors.
Data amounts exchanged by each node can span from few
bits to several million bits per second, depending on the sampling
frequency and the type of the acquired data. Therefore,
the total data traffic incoming from all the nodes can become
an important constraint for the network design and the optimization
of the communication protocol. Indeed, the cost of the
network can significantly contribute to the cost of the overall
system, especially when node locations are not fully covered
by the network.
Despite the fact that distributed systems allow one to partially
share resources, it is required that all the nodes embed
a proper communication interface able to support a common
protocol. Moreover, taking into account that protocols and
control software for a distributed system can be significantly
more complex than those found on a concentrated system, special
care should be taken in order to optimize the node design
and to minimize its cost. Often, node cost represents the most
significant part of the overall system cost, especially for those
systems featuring a large number of nodes.
Additionally, node power supply is another common limitation
in a distributed system. As a matter of fact, a suitable and
reliable power supply should be provided to each node. Unfortunately,
there are applications where, either due to node
locations difficult to reach or because of other constraints, it is
extremely difficult to provide a stable power supply for all the
nodes. In such situations, nodes should embed a sort of power
22
backup, as an example, based on rechargeable batteries, which
is able to provide a continuous power supply to all the nodes.
All of these aspects make design and maintenance of distributed
systems more complex and more expensive with respect
to concentrated instruments.
Eventually, other two aspects, which are often underestimated,
should be considered carefully: data security and
quality of measurement. Data acquired from the system can
be either sensitive or reserved, such as biomedical information
of patients and access to restricted areas. Such data should
be protected both from the access of unauthorized people and
from any possible alteration. Therefore, these applications
require suitable communication protocols and network infrastructures
that are designed to guarantee the required level of
data security.
Furthermore, acquired data can represent important information
which is necessary to know accurately and effectively,
as in the case of critical and life-related applications [2]. This
aspect is rarely discussed in the literature, and assuring the required
level of measurement accuracy for all nodes during all
the system operative life can be extremely difficult. Therefore,
this aspect should be carefully considered during the design of
the system. As an example, during the system development,
it is possible to foresee the possibility either to perform periodic
calibrations of the sensing nodes or to periodically receive
feedback on the current measurement accuracy. Preferably,
such procedures should work remotely to decrease the maintenance
cost, especially in those systems counting hundreds of
nodes placed in distant locations.
Nowadays, distributed measurement systems find applications
in several fields from scientific research to consumer
services. Their size and extension range from few meters up
to thousands kilometers, and they can feature from few tens of
nodes to several thousands. For instance, the Square Kilometre
Array (SKA) [3] is an international distributed radio telescope
which, even though still under construction, will feature hundreds
of thousand antennas. Approximately eight terabits
per second of data will be synchronously collected by these
independent antennas, mainly located in South Africa and
Australia, and processed together to achieve unprecedented
performance and resolution.
Despite such impressive systems, there are smaller distributed
systems which find application in every-day life. From
smart traffic light controllers and other smart city applications
[4], [5], to home automation [6], environmental monitoring [7],
[8] and health care [9], the applications of such systems are virtually
unlimited.
Sometimes, it is not possible to employ wired networks
either due to the necessity of avoiding wires or to mobility constraints.
In such situations, a Wireless Sensor Network (WSN)
can be employed with success. Such a solution allows one to
connect several sensing nodes by using a suitable radio link
as network infrastructure. The advantages of such technology
are relevant, and this explains why in recent decades, most of
the research carried out in such fields involves the development
of new technologies and protocols for WSNs.
IEEE Instrumentation & Measurement Magazine
June 2022
Instrumentation & Measurement Magazine 25-4

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 25-4
