162
E. Fraccaroli and D. Quaglia
distributed IoT application to formulate and solve the design problem by using an
optimization approach. Such formal specification considers the following aspects:
• Energy consumption and type of energy source. Energy consumption is proportional to the throughput of the node. Therefore, it can be reduced by lowering
the physical bitrate or introducing sleeping periods. High energy consumption
requires to connect the device to a stable energy source such as the power grid or
a large battery to be recharged periodically by a person. Devices that exhibit such
requirements cannot be unattended. This is the case of gateways, coordinators,
network servers, routers. Medium energy consumption allows the use of batteries
that have autonomy of at least 2–3 years. In this case, the replacement period is
acceptable and can be combined with other activities already required in that
application context. This is the case of beacons that highlight the presence of an
object (e.g., in a museum). Low energy consumption allows the use of energyharvesting (or energy scavenging) techniques to obtain a small amount of energy
from the surrounding environment [93]. Well-known examples are NFC passive
tags that use radio-frequency energy and sensors supplied by photovoltaic cells.
• Frequency band. There are two main sets of frequency bands, i.e., sub-GHz and
millimeter-wave. Sub-GHz bands allow low-power and long-range operation as
well as obstacle avoidance at the cost of data rates usually far below 1 Mb/s.
Millimeter-wave bands (e.g., 2.4 and 5 GHz bands) allow higher data rates but
requires “line-of-sight” transmission. It is worth noting that frequency bands
that are blocked by walls (as in the case of 5 GHz band) allow intrinsic security
with respect to eavesdropping. Furthermore, as reported below, the choice of an
unlicensed or licensed frequency standard has an impact on telecommunication
cost.
• Quality of service. This is the general term that encompasses some different
performance aspects of the communication. One aspect is the data rate, which
denotes the amount of information transferred in the time unit. Multimedia
data (e.g., the output of a camera) usually require a higher data rate than
physical sensing (e.g., light intensity). If data from several sensors are merged,
the resulting stream may require a high data rate. It is worth noting that data
processing at the edge of the network can help reduce the required data rate;
for instance, in a video surveillance application, object recognition performed by
the camera allows to replace video transmission with a simple alert notification.
Another aspect is delay which denotes the time to move a message from one node
to another of the network. Delay constraints are important in control applications.
The absolute value of delay affects the promptness in device actuation. In closedloop control applications, the delay variation is very important since it affects
the stability of control. Finally, the third aspect is error rate which denotes the
fraction of bits whose value is erroneously received at the destination. Bit errors,
if detected, decrease data rate since the message is discarded; otherwise, they can
compromise data processing at the destination.
• Transmission range. The right distance between two communicating devices
is an application requirement. For some applications, e.g., precision agriculture
E. Fraccaroli and D. Quaglia
distributed IoT application to formulate and solve the design problem by using an
optimization approach. Such formal specification considers the following aspects:
• Energy consumption and type of energy source. Energy consumption is proportional to the throughput of the node. Therefore, it can be reduced by lowering
the physical bitrate or introducing sleeping periods. High energy consumption
requires to connect the device to a stable energy source such as the power grid or
a large battery to be recharged periodically by a person. Devices that exhibit such
requirements cannot be unattended. This is the case of gateways, coordinators,
network servers, routers. Medium energy consumption allows the use of batteries
that have autonomy of at least 2–3 years. In this case, the replacement period is
acceptable and can be combined with other activities already required in that
application context. This is the case of beacons that highlight the presence of an
object (e.g., in a museum). Low energy consumption allows the use of energyharvesting (or energy scavenging) techniques to obtain a small amount of energy
from the surrounding environment [93]. Well-known examples are NFC passive
tags that use radio-frequency energy and sensors supplied by photovoltaic cells.
• Frequency band. There are two main sets of frequency bands, i.e., sub-GHz and
millimeter-wave. Sub-GHz bands allow low-power and long-range operation as
well as obstacle avoidance at the cost of data rates usually far below 1 Mb/s.
Millimeter-wave bands (e.g., 2.4 and 5 GHz bands) allow higher data rates but
requires “line-of-sight” transmission. It is worth noting that frequency bands
that are blocked by walls (as in the case of 5 GHz band) allow intrinsic security
with respect to eavesdropping. Furthermore, as reported below, the choice of an
unlicensed or licensed frequency standard has an impact on telecommunication
cost.
• Quality of service. This is the general term that encompasses some different
performance aspects of the communication. One aspect is the data rate, which
denotes the amount of information transferred in the time unit. Multimedia
data (e.g., the output of a camera) usually require a higher data rate than
physical sensing (e.g., light intensity). If data from several sensors are merged,
the resulting stream may require a high data rate. It is worth noting that data
processing at the edge of the network can help reduce the required data rate;
for instance, in a video surveillance application, object recognition performed by
the camera allows to replace video transmission with a simple alert notification.
Another aspect is delay which denotes the time to move a message from one node
to another of the network. Delay constraints are important in control applications.
The absolute value of delay affects the promptness in device actuation. In closedloop control applications, the delay variation is very important since it affects
the stability of control. Finally, the third aspect is error rate which denotes the
fraction of bits whose value is erroneously received at the destination. Bit errors,
if detected, decrease data rate since the message is discarded; otherwise, they can
compromise data processing at the destination.
• Transmission range. The right distance between two communicating devices
is an application requirement. For some applications, e.g., precision agriculture