iot_module4.pdf

MODULE 4
Data and Analytics for IoT

 As more and more devices are added to IoT networks, the
data generated by these systems becomes overwhelming
 Traditional data management systems are simply unprepared
for the demands of what has come to be known as “big data.”
 The real value of IoT is not just in connecting things but
rather in the data produced by those things, the new services you
can enable via those connected things, and the business insights
that the data can reveal.
 However, to be useful, the data needs to be handled in a way
that is organized and controlled.
 Thus, a new approach to data analytics is needed for the
Internet ofThings

An Introduction to Data Analytics for IoT
 In the world of IoT, the creation of massive amounts of data
from sensors is common and one of the biggest challenges—
not only from a transport perspective but also from a data
management standpoint
 Modern jet engines are fitted with thousands of sensors that
generate a whopping 10GB of data per second
 Analyzing this amount of data in the most efficient manner possible
falls under the umbrella of data analytics

 Not all data is the same; it can be categorized and thus
analyzed in different ways.
 Depending on how data is categorized, various data analytics
tools and processing methods can be applied.
 Two important categorizations from an IoT
perspective are whether the data is structured or unstructured
and whether it is in motion or at rest.

Structured Versus Unstructured Data
 Structured data and unstructured data are important
classifications as they typically require different toolsets from a
data analytics perspective
 Structured data means that the data follows a model or
schema that defines how the data is represented or organized,
meaning it fits well with a traditional relational database
management system (RDBMS).
 In many cases you will find structured data in a simple
tabular form—for example, a spreadsheet where data
occupies a specific cell and can be explicitly defined and referenced

 Structured data can be found in most computing systems
and includes everything from banking transaction and
invoices to computer log files and router configurations.
 IoT sensor data often uses structured values, such as
temperature, pressure, humidity, and so on, which are all
sent in a known format.
 Structured data is easily formatted,stored,queried,and processed

 Because of the highly organizational format of structured
data, a wide array of data analytics tools are readily
available for processing this type of data.
 From custom scripts to commercial software like Microsoft
Excel and Tableau

 Unstructured data lacks a logical schema for
understanding and decoding the data through traditional
programming means.
 Examples of this data type include text, speech, images,
and video.
 As a general rule, any data that does not fit neatly into a
predefined data model is classified as unstructured
data

 According to some estimates, around 80% of a business’s
data is unstructured.
 Because of this fact, data analytics methods that can be
applied to unstructured data, such as cognitive
computing and machine learning, are deservedly garnering a
lot of attention.
 With machine learning applications, such as natural
language processing (NLP), you can decode speech.
 With image/facial recognition applications, you can extract
critical information from still images and video

 Smart objects in IoT networks generate both
structured and unstructured data.
 Structured data is more easily managed and processed due to
its well-defined organization.
 On the other hand, unstructured data can be harder to deal
with and typically requires very different analytics tools for
processing the data

Data in Motion Versus Data at Rest
 Data in IoT networks is either in transit (“data in motion”) or
being held or stored (“data at rest”).
 Examples of data in motion include traditional client/server
exchanges,such as web browsing and file transfers,and email.
 Data saved to a hard drive, storage array, or USB drive is
data at rest.

 From an IoT perspective, the data from smart objects is
considered data in motion as it passes through the network en route
to its final destination.
 This is often processed at the edge,using fog computing.
 When data is processed at the edge, it may be filtered and deleted
or forwarded on for further processing and possible storage at a fog
node or in the data center.
 Data does not come to rest at the edge.
 When data arrives at the data center, it is possible to process it in
real-time, just like at the edge, while it is still in motion.
 Tools with this sort of capability, are Spark, Storm, and Flink

 Data at rest in IoT networks can be typically found in
IoT brokers or in some sort of storage array at the data
center
 Hadoop not only helps with data processing but also data
storage

IoT Data Analytics Overview
 The true importance of IoT data from smart objects
is realized only when the analysis of the data leads to
actionable business intelligence and insights.
 Data analysis is typically broken down by the types
of results that are produced

Types of Data Analysis Results

Four types of data analysis results
 Descriptive:
 Descriptive data analysis tells you what is happening, either
now or in the past.
 For example, a thermometer in a truck engine reports
temperature values every second.
 From a descriptive analysis perspective, you can pull this data at
any moment to gain insight into the current operating
condition of the truck engine.
 If the temperature value is too high, then there may be
a cooling problem or the engine may be experiencing too
much load.

 Diagnostic:
 When you are interested in the “why,” diagnostic data analysis
can provide the answer.
 Continuing with the example of the temperature sensor in the
truck engine, you might wonder why the truck engine failed.
 Diagnostic analysis might show that the temperature of
the engine was too high, and the engine overheated.
 Applying diagnostic analysis across the data generated by a wide
range of smart objects can provide a clear picture of why a
problem or an event occurred

 Predictive:
 Predictive analysis aims to foretell problems or issues
before they occur.
 For example, with historical values of temperatures for the truck
engine, predictive analysis could provide an estimate on
the remaining life of certain components in the engine.
 These components could then be proactively replaced before
failure occurs.
 Or perhaps if temperature values of the truck engine start to rise
slowly over time, this could indicate the need for an oil change
or some other sort of engine cooling maintenance.

 Prescriptive:
 Prescriptive analysis goes a step beyond predictive and recommends
solutions for upcoming problems.
 A prescriptive analysis of the temperature data from a truck engine
might calculate various alternatives to cost-effectively
maintain our truck
 These calculations could range from the cost necessary for more frequent oil
changes and cooling maintenance to installing new cooling equipment on the
engine or upgrading to a lease on a model with a more powerful engine.
 Prescriptive analysis looks at a variety of factors and makes the
appropriate recommendation

 Both predictive and prescriptive analyses are more resource
intensive and increase complexity, but the value they provide
is much greater than the value from descriptive and
diagnostic analysis

IoT Data Analytics Challenges
Problems by using RDMS in IoT
1. Scaling Problems (performance issues, costly to resolve,
req more h/w, architechture changes)
2. Volatility of Data (change in schema)

Machine Learning
 ML is central to IoT.
 Data collected by smart objects needs to be analyzed, and
intelligent actions need to be taken based on these analyses.
 Performing this kind of operation manually is almost impossible
(or very, very slow and inefficient).
 Machines are needed to process information fast and
react instantly when thresholds are met
 Ex: advances in self-driving vehicle--abnormal pattrn
recognition in a crowd and automated intelligent and
machine-assisted decision system

Machine Learning Overview
 Machine learning is, in fact, part of a larger set of technologies
commonly grouped under the term artificial intelligence (AI).
 AI includes any technology that allows a computing system to
mimic human intelligence using any technique, from very
advanced logic to basic “if-then-else” decision loops.
 Any computer that uses rules to make decisions belong to
this group

 A simple example is an app that can help you find
your parked car.
 A GPS reading of your position at regular intervals calculates
your speed.
 A basic threshold system determines whether you are driving
(for example, “if speed > 20 mph or 30 kmh, then start
calculating speed”).
 When you park and disconnect from the car Bluetooth
system, the app simply records the location when the
disconnection happens.
 This is where your car is parked.

 In more complex cases, static rules cannot be simply
inserted into the program because they require parameters that
can change or that are imperfectly understood
 A typical example is a dictation program that runs on a
computer.
 The program is configured to recognize the audio pattern
of each word in a dictionary,but it does not know your
voice’s specifics—your accent, tone, speed, and so on

 You need to record a set of predetermined sentences to
help the tool match well-known words to the sounds
you make when you say the words.
 This process is called machine learning.
 ML is concerned with any process where the
computer needs to receive a set of data that is
processed to help perform a task with more
efficiency.
 ML is a vast field but can be simply divided in two main
categories: supervised and unsupervised learning

Supervised Learning
 In supervised learning, the machine is trained with input for
which there is a known correct answer.
 For example, suppose that you are training a system to recognize
when there is a human in a mine tunnel.
 A sensor equipped with a basic camera can capture shapes
and return them to a computing system that is responsible
for determining whether the shape is a human or
something else (such as a vehicle, a pile of ore, a rock, a piece of
wood, and so on.).

 With supervised learning techniques, hundreds or thousands
of images are fed into the machine, and each image is labelled
(human or nonhuman in this case).
 This is called the training set.
 An algorithm is used to determine common parameters
and common differences between the images.
 The comparison is usually done at the scale of the entire
image, or pixel by pixel.
 Images are resized to have the same characteristics
(resolution, color depth, position of the central figure, and
so on), and each point is analyzed.

 Each new image is compared to the set of known “good images,” and a
deviation is calculated to determine how different, the new
image is from the average human image and, therefore, the
probability that what is shown is a human figure. This process is
called classification.
 After training, the machine should be able to recognize human shapes.
Before real field deployments, the machine is usually tested with
unlabeled pictures— this is called the validation or the test set,
depending on the ML system used—to verify that the recognition
level is at acceptable thresholds. If the machine does not reach the
level of success expected, more training is needed

 In other cases, the learning process is not about classifying in two
or more categories but about finding a correct value.
 For example, the speed of the flow of oil in a pipe is a
function of the size of the pipe,the viscosity of the oil,pressure,and a few
other factors.
 When you train the machine with measured values, the machine
can predict the speed of the flow for a new, and unmeasured,
viscosity.
 This process is called regression; regression predicts numeric
values,whereas classification predicts categories

Unsupervised Learning
 In some cases, supervised learning is not the best method for a
machine to help with a human decision.
 Suppose that you are processing IoT data from a factory
manufacturing small engines.
 You know that about 0.1% of the produced engines on average
need adjustments to prevent later defects, and your task is to
identify them before they get mounted into machines and shipped
away from the factory.
 With hundreds of parts, it may be very difficult to detect the
potential defects, and it is almost impossible to train a machine to
recognize issues that may not be visible

 However, you can test each engine and record multiple
parameters, such as sound, pressure, temperature of key
parts, and so on.
 Once data is recorded, you can graph these elements in
relation to one another (for example, temperature as a
function of pressure, sound versus rotating speed
overtime).
 You can then input this data into a computer and use
mathematical functions to find groups.

 For example, you may decide to group the engines by the
sound they make at a given temperature.
 A standard function to operate this grouping, K-means clustering,
finds the mean values for a group of engines (for example,mean
value for temperature,mean frequency for sound).
 Grouping the engines this way can quickly reveal several types of
engines that all belong to the same category (for example, small
engine of chainsaw type, medium engine of lawnmower type).
 All engines of the same type produce sounds and temperatures in the
same range as the other members of the same group.

 There will occasionally be an engine in the group that
displays unusual characteristics (slightly out of
expected temperature or sound range).
 This is the engine that you send for manual evaluation.
 The computing process associated with this determination is
called unsupervised learning.
 This type of learning is unsupervised because there is not a“good”
or“bad”answer known in advance.
 It is the variation from a group behavior that allows the
computer to learn that something is different

Neural Networks
 Processing multiple dimensions requires a lot of computing power.
 It is also difficult to determine what parameters to input and what
combined variations should raise red flags.
 Similarly, supervised learning is efficient only with a large training
set; larger training sets usually lead to higher accuracy in the
prediction
 Training the machines was often deemed too expensive and
complicated.

 Distinguishing between a human and a car is easy.
 The computer can recognize that humans have distinct shapes (such
as legs or arms) and that vehicles do not.
 But, distinguishing a human from another mammal is much more
difficult
 The same goes for telling the difference between a pickup truck
and a van.
 You can tell when you see one, but training a machine to
differentiate them requires more than basic shape recognition

 Neural networks are ML methods that mimic the way the human
brain works.
 When you look at a human figure, multiple zones of your brain are
activated to recognize colors, movements, facial expressions, and so
on.
 Your brain combines these elements to conclude that the shape you
are seeing is human
 Neural networks mimic the same logic.
 The information goes through different algorithms (called units),
each of which is in charge of processing an aspect of the
information

 The resulting value of one unit computation can be used
directly or fed into another unit for further processing to
occur
 For example, a neural network processing human image
recognition may have two units in a first layer that determines
whether the image has straight lines and sharp angles—because vehicles
commonly have straight lines and sharp angles, and human figures do not.
 If the image passes the first layer successfully (because there are no or
only a small percentage of sharp angles and straight lines), a
second layer may look for different features (presence of
face, arms, and so on), and then a third layer might
compare the image to images of various animals and
conclude that the shape is a human (or not)

 Neural networks rely on the idea that information is divided
into key components, and each component is assigned a
weight.
 The weights compared together decide the classification of this
information (no straight lines + face + smile = human).
 When the result of a layer is fed into another layer, the process is
called deep learning (“deep” because the learning process has
more than a single layer).
 One advantage of deep learning is that having more layers allows
for richer intermediate processing and representation of the
data.
 At each layer, the data can be formatted to be better utilized
by the next layer. This process increases the efficiency of
the overall result

Machine Learning and Getting Intelligence from Big Data
 ML operations can be organized into two broad subgroups:
 Local learning
 Data is collected and processed locally, either in the sensor itself
(the edge node) or in the gateway (the fog node)
 Remote learning
 Data is collected and sent to a central computing unit
(typically the data center in a specific location or in the
cloud), where it is processed.

 Regardless of the location where data is processed, common
applications of ML for IoT revolve around four major
domains:
 Monitoring
 Smart objects monitor the environment where they operate
 Example such as air temperature, humidity, or presence of
carbon dioxide in a mine etc

 Behavior control
 Monitoring commonly works in conjunction with behavior control.
 When a given set of parameters reach a target threshold —
defined in advance (that is, supervised) or learned
dynamically through deviation from mean values (that
is, unsupervised)—monitoring functions generate an
alarm.
 This alarm can be relayed to a human, but a more efficient and
more advanced system would trigger a corrective action,
such as increasing the flow of fresh air in the mine tunnel, turning the
robot arm, or reducing the oil pressure in the pipe.

 Operations optimization
 Behavior control typically aims at taking corrective actions
based on thresholds.
 However, analyzing data can also lead to changes that
improve the overall process
 For example, a water purification plant in a smart city
can implement a system to monitor the efficiency of
the purification process based on which chemical
(from company A or company B) is used,at what
temperature,and associated to what stirring mechanism
(stirring speed and depth).

 Self-healing, self-optimizing
 The ML engine can be programmed to dynamically
monitor and combine new parameters (randomly or
semi-randomly) and automatically deduce and
implement new optimizations when the results
demonstrate a possible gain.
 The system becomes self-learning and self optimizing
 It also detects new K-means deviations that result in
predetection of new potential defects, allowing the system to
self-heal

Predictive Analytics
 Machine learning and big data processing for IoT fit very well into th
digitization
 The advanced stages of this model see the network self-
diagnose and self-optimize.
 In the IoT world, this behavior is what the previous section describes
 When data from multiple systems is combined and analyzed
together,predictions can be made about the state of the system.
For example,
 case of sensors deployed on locomotives. Multiple smart objects
measure the pull between carriages, the weight on each wheel, and
multiple other parameters to offer a form of cruise control
optimization for the driver.

Predictive Analytics
 At the same time, cameras observe the state of the tracks
ahead, audio sensors analyze the sound of each wheel on the
tracks, and multiple engine parameters are measured and
analyzed.
 All this data can be returned to a data processing center in
the cloud that can re-create a virtual twin of each
locomotive.
 Modeling the state of each locomotive and combining this
knowledge with anticipated travel and with the states (and
detected failures) of all other locomotives of the same type
circulating on the tracks of the entire city, province, state, or
country allows the analytics platform to make very accurate
predictions on what issue is likely to affect each
train and each locomotive.

 Such predictive analysis allows preemptive
maintenance and increases the safety and efficiency
of operations.
 Similarly, sensors combined with big data can anticipate
defects or issues in vehicles operating in mines, in
manufacturing machines, or any system that can be
monitored, along with other similar systems.

Big Data Analytics Tools and Technology
 Big data analytics can consist of many different software
pieces that together collect,store,manipulate,and analyze
all different data types.
 Generally, the industry looks to the “three Vs”to categorize big
data:
 Velocity
 Refers to how quickly data is being collected and analyzed.
 Hadoop Distributed File System is designed to ingest and process
data very quickly.
 Smart objects can generate machine and sensor data at a very fast rate
and require database or file systems capable of equally fast ingest
functions.

 Variety
 refers to different types of data.
 Often you see data categorized as structured,semi-structured,or
unstructured.
 Different database technologies may only be capable of accepting one
of these types.
 Hadoop is able to collect and store all three types

 Volume
 refers to the scale of the data.
 Typically, this is measured from gigabytes on the very low
end to petabytes or even exabytes of data on the other
extreme

 The characteristics of big data can be defined by the
sources and types of data.
 First is machine data, which is generated by IoT devices and
is typically unstructured data.
 Second is transactional data, which is from sources that produce
data from transactions on these systems, and, have high volume and
structured.
 Third is social data sources, which are typically high volume
and structured.
 Fourth is enterprise data, which is data that is lower in
volume and very structured
 Hence big data consists of data from all these separate
sources.

Massively Parallel Processing Databases
 Enterprises have used relational databases for storing structured,
row and column style data types for decades.
 Relational databases are often grouped into a broad data storage
category called data warehouses.
 Though they are the centerpiece of most data architectures, they
are often used for longer-term archiving and data queries that can
often take minutes or hours
 Massively parallel processing (MPP) databases were
built on the concept of the relational data warehouses
but are designed to be much faster, to be efficient, and
to support reduced query times

 To accomplish this, MPP databases take advantage of multiple
nodes (computers) designed in a scale-out architecture
such that both data and processing are distributed across
multiple systems
 MPPs are sometimes referred to as analytic databases because
they are designed to allow for fast query processing and often have
built-in analytic functions
 As the name implies, these database types process massive data sets
in parallel across many processors and nodes

 An MPP architecture typically contains a single master node that is
responsible for the coordination of all the data storage and
processing across the cluster.
 It operates in a “shared-nothing” fashion, with each node
containing local processing, memory, and storage and operating
independently.

 Data storage is optimized across the nodes in a
structured SQL-like format that allows data analysts to work
with the data using common SQL tools and applications

NoSQL Databases
 NoSQL (“not only SQL”) is a class of databases that support
semi-structured and unstructured data, in addition to the
structured data handled by data warehouses and MPPs
 NoSQL is not a specific database technology; rather, it is an
umbrella term that encompasses several different types of
databases,including the following
 Document stores:
 This type of database stores semi-structured data, such as
XML or JSON.
 Document stores generally have query engines and indexing
features that allow for many optimized queries

 Key-value stores
This type of database stores associative arrays where a key
is paired with an associated value.These databases are
easy to build and easy to scale
 Wide-column stores
This type of database stores similar to a key-value store,
but the formatting of the values can vary from row to row,
even in the same table

 Graph stores
 This type of database is organized based on the relationships
between elements.
 Graph stores are commonly used for social media or natural
language processing, where the connections between data
are very relevant.

Hadoop
 Hadoop is the most recent entrant into the data management
market, but it is arguably the most popular choice as a data
repository and processing engine.
 Hadoop was originally developed as a result of projects at
Google andYahoo!
 The original intent for Hadoop was to index millions of
websites and quickly return search results for open source
search engines

 Initially, the project had two key elements:
 Hadoop Distributed File System (HDFS):
A system for storing data across multiple nodes
 MapReduce:
 A distributed processing engine that splits a large task into
smaller ones that can be run in parallel.
 Hadoop relies on a scale-out architecture that leverages local
processing,memory,and storage to distribute tasks and
provide a scalable storage system for data.

 Both MapReduce and HDFS take advantage of this
distributed architecture to store and process massive
amounts of data and are thus able to leverage resources
from all nodes in the cluster.
 For HDFS, this capability is handled by specialized nodes in
the cluster, including NameNodes and DataNodes

Fig; Distributed Hadoop Cluster

 NameNodes
 These are a critical piece in data adds, moves, deletes, and
reads on HDFS.
 They coordinate where the data is stored, and maintain a map
of where each block of data is stored and where it is
replicated.
 All interaction with HDFS is coordinated through the
primary (active) NameNode, with a secondary (standby)
NameNode notified of the changes in the event of a failure of
the primary.
 The NameNode takes write requests from clients and
distributes those files across the available nodes in
configurable block sizes, usually 64 MB or 128 MB blocks.
 The NameNode is also responsible for instructing the
DataNodes where replication should occur.

 DataNodes
 These are the servers where the data is stored at the direction
of the NameNode.
 It is common to have many DataNodes in a Hadoop cluster to
store the data.
 Data blocks are distributed across several nodes and often are
replicated three, four, or more times across nodes for
redundancy.
 Once data is written to one of the DataNodes, the DataNode
selects two (or more) additional nodes, based on replication
policies, to ensure data redundancy across the cluster

YARN
 Introduced with version 2.0 of Hadoop,YARN (YetAnother
Resource Negotiator) was designed to enhance the
functionality of MapReduce.
 With the initial release, MapReduce was responsible for
batch data processing and job tracking and resource
management across the cluster.
 YARN was developed to take over the resource negotiation
and job/task tracking, allowing MapReduce to be responsible
only for data processing.

The Hadoop Ecosystem
 Since the initial release of Hadoop in 2011, many projects
have been developed to add incremental functionality to
Hadoop and have collectively become known as the Hadoop
ecosystem.
 Apache Kafka
 Apache Spark
 Apache Storm and Apache Flink
 Lambda Architecture

Comparing Big Data and Edge Analytics
 When you hear the term big data,it is usually in reference to
unstructured data that has been collected and stored in the
cloud
 Tools like Hadoop and MapReduce are great at tackling
problems that require deep analytics on a large and complex
quantity of unstructured data;
 However, due to their distance from the IoT endpoints and
the bandwidth required to bring all the data back to the
cloud, they are generally not well suited to real-time analysis
of data as it is generated.

 In applying data analytics to the car racing example, big data
analytics is used to examine all the statistics of the racing
team and players based on their performance in the data
center or cloud
 Streaming analytics involves analyzing a race while it is
happening and trying to figure out who is going to win based
on the actual performance in real-time—and this analysis is
typically performed as close to the edge as possible.
 Streaming analytics allows you to continually monitor and
assess data in real-time so that you can adjust or fine-tune
your predictions as the race progresses.

 In the context of IoT, with streaming analytics performed at
the edge (either at the sensors themselves or very close to
them, in a fog node that is, for example, integrated into the
gateway), it is possible to process and act on the data in
realtime without waiting for the results from a future batch-
processing job in the cloud.

 The key values of edge streaming analytics include the
following:
 Reducing data at the edge
 Analysis and response at the edge
 Time sensitivity

Edge Analytics Core Functions
 To perform analytics at the edge, data needs to be viewed as
real-time flows.
 Whereas big data analytics is focused on large quantities of
data at rest, edge analytics continually processes streaming
flows of data in motion
 Streaming analytics at the edge can be broken down into
three simple stages:
 Raw input data
 Analytics processing unit (APU)
 Output streams

 In order to perform analysis in real-time, the APU needs to
perform the following functions:
 Filter
 Transform
 Time
 Correlate

 Match patterns
 Improve business intelligence

Distributed Analytics Systems
 Depending on the application and network architecture,
analytics can happen at any point throughout the IoT system.
 Streaming analytics may be performed directly at the edge, in
the fog, or in the cloud data center.
 There are no hard and- fast rules dictating where analytics
should be done, but there are a few guiding principles
 Sometimes better insights can be gained and data responded
to more intelligently when we step back from the edge and
look at a wider data set.
 This is the value of fog computing.

 Fog analytics allows you to see beyond one device, giving you
visibility into an aggregation of edge nodes and allowing you
to correlate data from a wider set
 Example of an oil drilling company that is measuring both
pressure and temperature on an oil rig

Network Analytics
 Another form of analytics that is extremely important in
managing IoT systems is network-based analytics
 Network analytics is concerned with discovering patterns in
the communication flows from a network traffic perspective.
 Network analytics has the power to analyze details of
communications patterns made by protocols and correlate
this across the network.
 It allows you to understand what should be considered
normal behavior in a network and to quickly identify
anomalies that suggest network problems due to suboptimal
paths, intrusive malware, or excessive congestion.

 Information technology (IT) environments have faced active
attacks and information security threats for many decades,
and the incidents and lessons learned are well-known and
documented.
 Operational technology (OT) environments were
traditionally kept in silos and had only limited connection to
other networks.
 Thus, the history of cyber attacks on OT systems is much
shorter and has far fewer incidents documented
 Security in the OT world also addresses a wider scope than in
the IT world. For example, in OT, the word security is almost
synonymous with safety

 A Brief History of OT Security
 Common Challenges in OT Security
 How IT and OT Security Practices and SystemsVary
 Formal Risk Analysis Structures: OCTAVE and FAIR
 The PhasedApplication of Security in an Operational
Environment

A Brief History of OT Security
 Cybersecurity incidents in industrial environments can result
in physical consequences that can cause threats to human lives
as well as damage to equipment, infrastructure, and the
environment.
 While there are certainly traditional IT-related security
threats in industrial environments, it is the physical
manifestations and impacts of the OT security incidents that
capture media attention and elicit broad-based public
concern.

 Historically, attackers were skilled individuals with deep
knowledge of technology and the systems they were attacking.
 However, as technology has advanced, tools have been created to
make attacks much easier to carry out.
 To further complicate matters, these tools have become more
broadly available and more easily obtainable.
 Compounding this problem, many of the legacy protocols used in
IoT environments are many decades old, and there was no thought
of security when they were first developed.
 This means that attackers with limited or no technical capabilities
now have the potential to launch cyber attacks, greatly increasing
the frequency of attacks and the overall threat to end operators.

Common Challenges in OT Security
 Erosion of Network Architecture
 Two of the major challenges in securing industrial
environments have been initial design and ongoing
maintenance.
 The initial design challenges arose from the concept that
networks were safe due to physical separation from the
enterprise with minimal or no connectivity to the outside
world, and the assumption that attackers lacked sufficient
knowledge to carry out security attacks.

 In many cases, the initial network design is sound and even
follows well-defined industrial best practices and standards
 The challenge, and the biggest threat to network security, is
standards and best practices either being misunderstood or
the network being poorly maintained.
 It is more common that, over time, what may have been a
solid design to begin with is eroded through ad hoc updates
and individual changes to hardware and machinery without
consideration for the broader network impact

Pervasive Legacy Systems
 Due to the static nature and long lifecycles of equipment in
industrial environments, many operational systems may be
deemed legacy systems
 In many cases, legacy components are not restricted to
isolated network segments but have now been consolidated
into the IT operational environment.
 From a security perspective, this is potentially dangerous as
many devices may have historical vulnerabilities or
weaknesses that have not been patched and updated

 Beyond the endpoints, the communication infrastructure and
shared centralized compute resources are often not built to
comply with modern standards.
 In fact, their communication methods and protocols may be
generations old and must be interoperable with the oldest
operating entity in the communications path.
 This includes switches, routers, firewalls, wireless access
points, servers, remote access systems, patch management,
and network management tools.
 All of these may have exploitable vulnerabilities and must be
protected

Insecure Operational Protocols
 Many industrial control protocols, were designed without
inherent strong security requirements
 Furthermore, their operation was often within an assumed
secure network.
 In addition to any inherent weaknesses or vulnerabilities,
their operational environment may not have been designed
with secured access control in mind

 Industrial protocols, such as supervisory control and data
acquisition (SCADA) ,particularly the older variants, suffer
from common security issues.
 Three examples of this are, lack of authentication between
communication endpoints, no means of securing and
protecting data at rest or in motion, and insufficient
granularity of control to properly specify recipients or avoid
default broadcast approaches.

 The structure and operation of most of these protocols is
often publicly available.
 While they may have been originated by a private firm, for
the sake of interoperability, they are typically published for
others to implement.
 Thus, it becomes a relatively simple matter to compromise
the protocols themselves and introduce malicious actors that
may use them to compromise control systems

Some common industrial protocols and their respective security concerns
 Modbus
 Modbus is commonly found in many industries, such as
utilities and manufacturing environments, and has multiple
variants (for example, serial,TCP/IP).
 It was created by the first programmable logic controller
(PLC) vendor, Modicon, and has been in use since the 1970s.
 It is one of the most widely used protocols in industrial
deployments, and its development is governed by the
Modbus Organization.

 Authentication of communicating endpoints was not a default
operation
 Some older and serial-based versions of Modbus
communicate via broadcast.
 The ability to curb the broadcast function does not exist in
some versions.There is potential for a recipient to act on a
command that was not specifically targeting it
 Validation of the Modbus message content is also not
performed by the initiating application

DNP3 (Distributed Network Protocol)
 DNP3 has placed great emphasis on the reliable delivery of
messages
 In the case of DNP3, participants allow for unsolicited
responses, which could trigger an undesired response.
 The missing security element here is the ability to establish
trust in the system’s state and thus the ability to trust the
veracity of the information being presented

ICCP (Inter-Control Center Communications Protocol)
 Initial versions of ICCP had several significant gaps in the
area of security.
 One key vulnerability is that the system did not require
authentication for communication.
 Second, encryption across the protocol was not enabled as a
default condition, thus exposing connections to man-in-the-
middle (MITM) and replay attacks

International Electrotechnical Commission (IEC) Protocols
 Three message types were initially defined: MMS
(Manufacturing Message Specification), GOOSE (Generic
Object Oriented Substation Event), and SV (Sampled
Values).
 Both GOOSE and SV operate via a publisher/subscriber
model, with no reliability mechanism to ensure that data has
been received
 Authentication is embedded in MMS, but it is based on clear-
text passwords, and authentication is not available in GOOSE
or SV

Device Insecurity
 Beyond the communications protocols that are used and the
installation base of legacy systems, control and communication
elements themselves have a history of vulnerabilities
 It is not difficult to understand why such systems are frequently
found vulnerable.
 First, many of the systems utilize software packages that can be
easily downloaded and worked against.
 Second, they operate on common hardware and standard
operating systems, such as MicrosoftWindows.
 Third,Windows and the components used within those
applications are well known to traditionally IT-focused security
researchers,There is little need to develop new tools or
techniques

Dependence on External Vendors
 Modern IT environments may be outsourcing business
operations or relegating certain processing or storage
functions to the cloud

How IT and OT Security Practices and Systems Vary
 The differences between an enterprise IT environment and an
industrial-focused OT deployment are important to understand
because they have a direct impact on the security practice applied
to them
 The Purdue Model for Control Hierarchy
 Regardless of where a security threat arises, it must be
consistently and unequivocally treated
 IT information is typically used to make business decisions, such as
those in process optimization, whereas OT information is instead
characteristically leveraged to make physical decisions, such as
closing a valve, increasing pressure, and so on

 The operational domain must also address physical safety and
environmental factors as part of its security strategy —and
this is not normally associated with the IT domain

The Purdue Model for Control Hierarchy

OT Network Characteristics Impacting Security
 While IT and OT networks are beginning to converge, they
still maintain many divergent characteristics in terms of how
they operate and the traffic they handle.
 These differences influence how they are treated in the
context of a security strategy.
 IT networks
 traffic traverse far
 They frequently traverse the network through layers of
switches and eventually make their way to a set of local or
remote servers, which they may connect to directly

 OT networks
 By comparison, in an OT environment (Levels 0–3), there
are typically two types of operational traffic.
 The first is local traffic that may be contained within a
specific package or area to provide local monitoring and
closed-loop control.
 This is the traffic that is used for real time (or near-real-time)
processes and does not need to leave the process control
levels.
 The second type of traffic is used for monitoring and control
of areas or zones or the overall system.

iot_module4.pdf

More Related Content

Similar to iot_module4.pdf (20)

Recently uploaded (16)

iot_module4.pdf