3 Engineering IoT Networks
163
and smart metering, long distance values are desirable to avoid the need for
intermediate systems to re-launch the signal. Vice versa, in proximity services
and electronic payments, keeping the maximum distance very short is a requirement. It is worth noting that large distance values increase power consumption,
electromagnetic noise, and risk of eavesdropping.
• Mobility. There are two kinds of mobility. In case of low mobility, users are
interested in avoiding the presence of cables to reduce cost or to be able to
change the position of nodes if needed. In the case of high mobility, devices
are transmitting while moving. This feature requires to adapt the transmission
to the changing environmental condition, e.g., distance from the counterpart,
reflections over obstacles, and presence of other electromagnetic sources in the
same frequency range.
• Scalability. Except for mesh networks, usual IoT infrastructures are based on
star or tree topology with intermediate systems connecting a set of end nodes.
In these cases the maximum number of end nodes supported by an intermediate
system is very important. Small values indicate low scalability of the technology
which leads to high cost to deploy intermediate systems.
• Security. Data confidentiality and integrity will become ever more important
in IoT applications according to their diffusion. Security at Data Link and
Network Layer should be directly enforced by the protocol standard. Security at
Application Layer can be easily provided by recurring to the traditional TCPbased mechanisms (e.g., TLS/SSL). If TCP is not supported by the protocol
stack, then custom mechanisms should be implemented by the application
designer.
• Cost. The cost of an IoT network infrastructure depends on different aspects.
The most evident aspect is the hardware cost of the various devices involved
in the application. Most of these devices are end nodes (e.g., sensors), but the
number of intermediate systems (e.g., gateways) can be significant especially if
the technology is not scalable. We can include in this category also the cost for
the use of cloud services. The second aspect is the telecommunication cost due to
subscription fees for the use of licensed frequency standards. The third aspect is
the energy cost, which encompasses energy bill, purchase of new batteries, and
disposal of the exhausted ones, as well as personnel cost for battery replacement.
Table 3.6 compares some of the standard technologies described before according to such aspects. It is interesting the comparison between standards based on
unlicensed frequencies (i.e., IEEE 802.15.4, Bluetooth, ZigBee, WirelessHART, ZWave, IEEE 802.11, LoRaWAN and Sigfox) and cellular standards (i.e., NB-IoT
and LTE-M). Regarding the first set, the use of the radio channel is free of charge but
the noise level due to its shared nature leads to a decrease of the QoS and scalability.
To cope with crowded channels and provide a wide coverage, more gateways are
needed thus increasing deployment cost. Vice versa, in cellular networks, end nodes
can be directly connected to the telecom infrastructure which natively guarantees
a higher QoS and coverage (e.g., for NB-IoT) at the cost of subscription fees. The
163
and smart metering, long distance values are desirable to avoid the need for
intermediate systems to re-launch the signal. Vice versa, in proximity services
and electronic payments, keeping the maximum distance very short is a requirement. It is worth noting that large distance values increase power consumption,
electromagnetic noise, and risk of eavesdropping.
• Mobility. There are two kinds of mobility. In case of low mobility, users are
interested in avoiding the presence of cables to reduce cost or to be able to
change the position of nodes if needed. In the case of high mobility, devices
are transmitting while moving. This feature requires to adapt the transmission
to the changing environmental condition, e.g., distance from the counterpart,
reflections over obstacles, and presence of other electromagnetic sources in the
same frequency range.
• Scalability. Except for mesh networks, usual IoT infrastructures are based on
star or tree topology with intermediate systems connecting a set of end nodes.
In these cases the maximum number of end nodes supported by an intermediate
system is very important. Small values indicate low scalability of the technology
which leads to high cost to deploy intermediate systems.
• Security. Data confidentiality and integrity will become ever more important
in IoT applications according to their diffusion. Security at Data Link and
Network Layer should be directly enforced by the protocol standard. Security at
Application Layer can be easily provided by recurring to the traditional TCPbased mechanisms (e.g., TLS/SSL). If TCP is not supported by the protocol
stack, then custom mechanisms should be implemented by the application
designer.
• Cost. The cost of an IoT network infrastructure depends on different aspects.
The most evident aspect is the hardware cost of the various devices involved
in the application. Most of these devices are end nodes (e.g., sensors), but the
number of intermediate systems (e.g., gateways) can be significant especially if
the technology is not scalable. We can include in this category also the cost for
the use of cloud services. The second aspect is the telecommunication cost due to
subscription fees for the use of licensed frequency standards. The third aspect is
the energy cost, which encompasses energy bill, purchase of new batteries, and
disposal of the exhausted ones, as well as personnel cost for battery replacement.
Table 3.6 compares some of the standard technologies described before according to such aspects. It is interesting the comparison between standards based on
unlicensed frequencies (i.e., IEEE 802.15.4, Bluetooth, ZigBee, WirelessHART, ZWave, IEEE 802.11, LoRaWAN and Sigfox) and cellular standards (i.e., NB-IoT
and LTE-M). Regarding the first set, the use of the radio channel is free of charge but
the noise level due to its shared nature leads to a decrease of the QoS and scalability.
To cope with crowded channels and provide a wide coverage, more gateways are
needed thus increasing deployment cost. Vice versa, in cellular networks, end nodes
can be directly connected to the telecom infrastructure which natively guarantees
a higher QoS and coverage (e.g., for NB-IoT) at the cost of subscription fees. The