Internet of Things

Module 2
Syllabus

Choice 1 (20 marks) Choice 2 (20 marks) Important (Priority

wise)

Core IoT functional stack Layer 3 (Applications and IoT Data Management
Layer 1 (Sensors and Actuators) Analytics) and Compute stack
Layer 2 (Communications sublayer) Analytics vs Control
Access network sublayer Data vs Network
Gateways and backhaul sublayer Analytics
Network transport sublayer IoT Data Management and
IoT Network management Compute stack
Layer 3 (Applications and Analytics)
Analytics vs Control
Data vs Network Analytics

The Core IoT functional stack

IoT networks are built around the concept of “smart objects” which use contextual information
and configured goals to perform actions. These actions can be self-contained, but in most
cases, they are not as smart objects report information to an external platform, wherein analysis
will be carried out.
From an architectural standpoint, the Core IoT functional stack is divided into 3 layers:
1. Things Layer
2. Communications network Layer
a. Access network sublayer
b. Gateways and backhaul network sublayer
c. Network transport sublayer
d. IoT network management sublayer
3. Application and Analytics layer
Layer 1: Things Layer (Sensors and Actuators layer):
At this layer, the physical devices need to fit the constraints of the environment in which they
are deployed while still being able to collect and provide the information needed. Eg: A sensor
placed inside a heat furnace needs to withstand extreme heat conditions, A sensor that is used
in underwater imaging should be water resistant, and so on. In brief, the physical devices need
to withstand the wear and tear of the environment in which they will be deployed.
There are myriad ways to classify smart objects (RRRBMO): (RRR + Bheem + O)
● Battery powered or Power connected: This classification is based on whether the
object carries its own energy supply or receives continuous power from an external
source. Battery-powered objects can be moved easily, but are limited with respect to
transmission range and size.
● Mobile or static: This classification is based on whether the object needs to constantly
move around or be stationed in one place. The mobility can be due to the object itself
moving around or because the deployed environment is mobile. Frequency and range are
critical parameters to be considered here.
● Low or high Reporting frequency: This classification is based on how often a smart
object reports the monitored parameters. Eg: A rust-sensing smart object which
reports the collected data once a month has a low reporting frequency, whereas a smart
traffic sensor may need to report the real-time traffic data on a minute-by-minute basis,
thus high reporting frequency. Higher frequencies drive higher energy consumption,
which may create constraints on the possible power source, object mobility, and
transmission range.
● Simple or Rich data: This classification is based on the nature and quantity of data
exchanged at each reporting cycle. Eg: Humidity sensor - Simple data; Engine
sensor - Rich data. Rich data drives higher energy consumption, which may create
constraints on the possible power source, object mobility, and transmission range.
Reporting frequency and nature of data are combined to define throughput.
● Range: This classification is based on the distance at which the gateway is located.
The range is mostly determined by Eg: Fitness band (object) and Phone (Gateway) - The
object needs to be at a visual distance from the gateway so that reported data can be
accessed. Here, even if we enable a longer range, it will be an unnecessary luxury as data
 from the fitness band cannot be accessed if the phone is not at a visual range. On the
other hand, consider a scenario where traffic congestion happening at a faraway place
needs to be reported to a smart vehicle so that rerouting can be done. Now the range
required would be several hundred meters or kilometers away.
● Object density per cell: This classification is based on the number of smart objects
connected to the same gateway over a given area. Eg: A single smart traffic sensor
monitoring the traffic in a not-so-busy junction can be considered low density whereas a
group of traffic sensors monitoring the traffic in a busy traffic junction is comparatively
high density.
Sensor applications based on Mobility and throughput
● Low Mobility, Low throughput: Industrial devices, Home security, and fire sensors
● Low Mobility, High throughput: Traffic cameras, Telemedicine
● High Mobility, Low throughput: Battlefield communications, Fleet management
● High Mobility, High throughput: Smart vehicle sensors, Video surveillance

Layer 2: Communications Network Layer

While designing the Communications Network layer, we will have to consider the fact that
computing and network assets used in IoT can be very different from those in traditional
IT and OT environments, such as differences in physical forms of devices used in IT and OT,
along with operational differences. The IoT environment demands that we take into account
certain factors such as suitability, reliability, while corresponding hardware devices are
designed. The hardware should able to handle temperature variances (extreme heat to extreme
cold), humidity fluctuations, operating extremes related to kinetic forces (low amplitude
constant vibrations vs. sudden acceleration or deceleration), the impact of solid particulates
(dust build-up due to the use of cooling fans), hazardous location design (corrosive nature of
the deployed environment), Impact equipment has on the environment (spark suppression design
in a chemical plant), and so on. OT systems also demand that redundant power supplies be built
into the device itself.
When smart objects are not self-contained, they need to communicate with an external
system. The Communications network layer is the link layer that acts as a bridge between the
Things and Application layers. It is further divided into 4 sublayers as follows:
Access network sublayer: The last mile of the IoT network is the access network. This is
typically made up of wireless technologies. Herein we will have to choose the type of access
technology used for communication purposes. There are several parameters that need to be
considered while opting for a particular access technology, such as Range, Throughput, Power
consumption, Cost, Network topology, Cell size, and so on.
Range: Based on range, one can choose from the following access technologies:
● PAN (Personal Area Network): Scale of a few meters; Personal space around a person;
Eg: Bluetooth
● HAN (Home Area Network): Scale of a few tens of meters; Eg: Zigbee, BLE
● NAN (Neighbourhood Area Network): Scale of a few hundreds of meters; A NAN is
generally installed by an individual to serve a family or a number of neighbors; Eg:
Mobile Hotspots, WLAN
● FAN (Field Area Network): Scale of several tens of meters to several hundred meters;
FAN is often seen as open space, and hence not secured and not controlled. FAN is
sometimes viewed as a group of NANs
● LAN (Local Area Network): Scale of up to 100 meters
● MAN (Metropolitan Area Network): Range of up to a few kilometers
● WAN (Wide Area Network): Range of more than few kilometers
● All these ranges are inclusive in nature
● Each protocol uses a specific frame format and transmission technique over a specific
frequency or band with varying throughput and range of transmission
● Increasing the throughput and distance demands more power. Therefore, cell size,
i.e.., the target quantity of objects in a single collection cell is an important parameter in
choosing an access technology
● Cost is also a determining factor in choosing an access technology
● Figure shown below combines cost, range (<10m, <1k, >1k), power consumption
(>500mA, >50mA, <50mA) and typical available bandwidth (<100kbps, <1Mbps,
<10Mbps, >10Mbps) for common IoT access technologies
● Network topology can be either Point-to-Point topologies or Point-to-multipoint
topologies
● Point-to-point topologies: These topologies allow one point to communicate with
another point (physical topology), which is uncommon in IoT use cases. Point-to-point
can also be when each object establishes an individual session with the gateway
(communication structure).
● Point-to-multipoint topologies: These topologies allow one point to communicate
with more than one other point.
● To form a network, a device needs to connect with another device. When both devices
fully implement the protocol stack functions, they can form a peer-to-peer network.
For example, in a house, temperature sensors may be deployed in each room, and they
may communicate with a central point where the temperature is displayed and controlled.
A room sensor does not need to communicate with another room sensor. Hence, the
control point is at the center of the network which forms a star topology, wherein the
control point is at the center (hub), and sensors are at the spokes. The central point
is called a coordinator/ full-function device (FFD). The sensor is called a
 reduced-function device (RFD). RFD cannot implement direct communications with
another RFD. There exists a master/ slave relationship between FFD and RFD.
● Topologies, where each FFD has a unique path to another FFD, are called cluster tree
topologies. FFDs in the cluster tree may have RFDs, resulting in a cluster star topology.
● If a node has more than one path to another node, it is called a mesh topology. This
communication can be used to directly exchange information between nodes or to
extend the range of the communication link. In this case, an intermediate node acts as
a relay between two other nodes, barring which the power and modulation constraints
make it impossible to establish communication. Range extension typically comes at the
price of slower communications. Eg: Wi-Fi mesh. Mesh networks exhibit
redundancy. The disappearance of one node does not necessarily interrupt network
communications. Data may still be relayed through other nodes to reach the intended
destination.
● Figure below shows mesh topology. Nodes A & D are too far apart to communicate
directly. In this case, communication can be relayed through nodes B or C. Node B may
be used as the primary relay. However the loss of node B does not prevent the
communication between nodes A & D as communication can take place via Node C.
Gateways and Backhaul Sublayer: Data collected from a smart object may need to be
forwarded to a central station where data is processed. As this station is often in a different
location from the smart object, data directly received from the sensor through an access
technology needs to be forwarded to another medium (the backhaul) and transported to the
central station. The gateway is in charge of this inter-medium connection. The role of the
gateway is to forward the collected information through a longer-range medium called the
backhaul to a headend central station where the information is processed. In most cases, the
smart objects are static or mobile, whereas the gateway is often static. Communication
between the sensors and the gateway may be wired or wireless. The choice of a backhaul
technology depends on the communication distance and also on the amount of data that needs
to be forwarded. When the smart object’s operation is controlled from a local site, and when the
environment is stable, Ethernet can be used as a backhaul. In unstable or changing
environments, wireless technology is used. Wi-Fi is common in this case, often with multiple
hops between the sensor field and the operation center. However throughput decreases as
node-to-node distance increases, and it also decreases with increase in the number of hops;
throughput halves for each additional hop.
WiMAX is an example of a longer-range technology (50km@70Mbps). Cellular technology is
also an option. There is always a trade-off between maximum rate and maximum range.
There is also the matter of operating in licensed vs. unlicensed bands. Licensed bands are more
reliable (less interference), but involve payment of usability fees. The choice of WiMAX or a
cellular technology depends on the vertical and the location.
Architectural considerations for WiMAX and Cellular Technologies

Technology Type and Architectural Characteristics (range, reliability,

Range adoption, power, cost)

Ethernet Wired, 100 m Requires a cable/ sensor; limited range but reliable, adapted
 max to static sensor environment

Wi-Fi (2.4 Wireless, 100m Can connect multiple clients (up to 200); limited range;
GHz, 5 to a few power consuming; large bandwidth; interference likely
GHz) kilometers

802.11ah Wireless, (1.5 - Can connect large number of clients (up to 6000); longer
(Sub-1 10) km range than traditional Wi-Fi; power efficient; limited
GHz) bandwidth; low adoption; cost may be an issue

WiMAX Wireless, up to Can connect large number of clients; Both licensed and
(802.16) 50 km unlicensed spectrum

Cellular Wireless, Can connect large number of clients; Large bandwidth

(LTE) Several available; Licensed spectrum - interference free, have to pay
Kilometers usability fees

Network Transport layer: For communication to be successful, network and transport

layer protocols such as IP and UDP must be implemented to support the variety of devices.
The network protocol needs to be open and standard-based to accommodate multiple
industries and multiple media. Scalability and security are also common requirements. IP is a
protocol that matches all these requirements. The flexibility of IP allows it to be embedded in
objects of very different natures, exchanging information over very different media.
Finally the transport layer protocols built above IP (UDP and TCP) can easily be leveraged
to decide whether the network should control the data packet delivery (TCP) or whether it
should be left to the application (UDP). UDP is a much lighter and faster protocol than TCP,
but does not guarantee packet delivery.
The transport structure requirement depends on the type of application:
1. Low mobility low throughput structure - IoT for the energy grid - 802.11ah, LPWA
2. Vertical push model and Downstream pull model
3. Unidirectional vs Bidirectional
4. Vertical vs. Mesh structure: Distribution automation (DA) allows energy meters to
communicate with each other so that load balancing can be optimized. With DA,
neighbor’s AC starts pulsing when your system pauses; this way air is kept fresh in both
the houses while the network is also stable instead of spiking up and down with
uncoordinated start and stop points.
 5. Bottom-up data transport flow vs. Top-down data flows
The communication structure may involve:
1. Peer-Peer
2. Point-Point
3. Point-Multipoint
4. Unicast and Multicast
IoT Network Management Sublayer: Additional protocols must be in place to allow the
headend applications to exchange data with the sensors
Some common IoT protocols used are:
HTTP: HTTP has a client and server component. The sensor could use the client part to
establish a connection to the IoT central application (the server). Limitation of HTTP is that it is
not suitable for constrained environments.
XMPP: Extensible Messaging and Presence protocol is based on instant messaging and
presence. It allows exchange of data between 2 or more systems. Limitation of XMPP is its
reliance on TCP which may force subscribers to maintain open sessions, thus a limitation for
memory constrained objects.
CoAP: Constrained Application Protocol is almost similar to HTTP, but has a smaller
header. CoAP runs on UDP and supports observation .
MQTT: Message Queue Telemetry Transport protocol uses a broker-based architecture. The
sensor can be set to be an MQTT publisher, the application that needs to receive the
information can be set as the MQTT subscriber, and any intermediary system can be set as a
broker to relay the information between the publisher and the subscriber/s. MQTT runs over
TCP because of which an MQTT client typically holds a connection open to the broker at all
times. This is a limiting factor for constrained environments.

Layer 3: Applications and Analytics Layer

Analytics Versus Control Applications
Analytics application: This type of application collects data from multiple smart objects,
processes the collected data, and displays information resulting from the data that was
processed. The processing is done so as to convey a view of the network that cannot be obtained
from solely looking at the information displayed by a single object.
Control application: This type of application controls the behavior of the smart object or an
object related to the smart object. For example, a pressure sensor connected to a pump. A
control application increases the speed when the connected sensor detects a drop in pressure.
Control applications are very useful in controlling complex aspects of an IoT network with a
logic that cannot be programmed inside a single IoT object, either because of complexity, or
elements being outside the IoT object. Eg: SCADA
Many advanced applications include both analytics and control modules. In most cases, data is
collected from the smart objects and processed in the analytics module. The result of the
processing may be used to modify the behavior of smart objects. The control module is used to
convey the instructions for behavioral changes.
Data Versus Network Analytics
Analytics is a general term that describes processing information to make sense of collected data
Data Analytics: This type of analytics processes the data collected by smart objects and
combines it to provide an intelligent view. At a very basic level, a dashboard can display an
alarm when a weight sensor detects that a shelf is empty in a store. In a more complex case,
temperature, pressure, wind, humidity, and light levels collected from thousands of sensors may
be combined and then processed to determine the likelihood of a storm and its possible path.
Data analytics can also monitor the IoT system itself. For example, a machine or robot in a
factory can report data about its own movements. This data can be used by an analytics
application to report degradation in the movement speeds, which may be indicative of a need to
service the robot before a part breaks.
Network Analytics: Most IoT systems are built around smart objects connected to the
network. A loss or degradation in connectivity is likely to affect the efficiency of the system.
For example, in open mines loss of connectivity may result in accidents of dump trucks. Loss of
connectivity also means data is no more flowing to the system, and thus system stops making
intelligent decisions. The control module is no more able to dictate behavioral changes in
smart objects.
Most analytics applications employ both data and network analytics modules. Network analytics
is necessary for connected systems, but the depth of analysis is subjective.
Data Analytics Versus Business Benefits
Data Analytics presents huge value in form of business benefits in IoT systems by collecting and
interpreting the data to arrive at intelligent decisions.
Static vs. Flexible IoT networks: IoT networks can be static wherein a clear list of elements to
monitor and analytics to perform are determined. Such static systems are common in
industrial environments where it is all about providing a clear view of the state of the operation.
Flexible IoT networks are open systems where the network is engineered to be flexible enough
that other sensors may be added in the future, and where both upstream and downstream
operations are allowed. This adds enormous value to the business.
Eg: Cisco Jasper - Consider the case of vending machines deployed throughout the city. At a
basic level, these machines can be connected, and sensors can be deployed to report when a
machine is in an error state. A repair person can be sent to address the issue when such a state is
identified. This saves time and reduces time to service. Now, with a flexible platform like Cisco
Jasper, new applications may be developed over time. For example, the machine sensors can be
improved to also report when an item is sold. This information can then be analyzed to get
valuable business insights such as which item is in demand and in which location, and so on.
This drives optimization.

Smart Services
The ability to use IoT to improve operations is often termed “smart services”. Smart
services often add an additional level of intelligence. Smart services use IoT and aim for
efficiency.
1. Sensors can be installed on equipment to ensure conformance with regulations - Weight
sensors
2. Hospitality and store management can be made smart by deploying occupancy
sensors. Additional personnel can be sent if needed based on the sensor’s output.
3. Sensors can be integrated into a light bulb, thus automatically switching on or off
depending on the occupancy status of the room. An even more advanced system can
study behavioral patterns, combine with other sensors to predict the arrival of a person,
and switch on the light before the arrival.
4. Smart grid applications can coordinate the energy consumption between houses to
regulate the energy demand from the grid. Power-draining applications such as washing
machines can be automatically made to operate when there is less energy demand.
5. In mining environments, vehicles can communicate to make sure that a dump truck is
always available
6. Smart traffic management systems can anticipate traffic jams and automatically reroute
public and private transport to avoid congestion