Lecture 14

IIT Kanpur July 201
20 followers
Mar 3, 2024
Edge Computing security and privacy I am
Dr Raju mistra Professor Department of
computer science and engineering Indian
Institute of Technology Patna contents
we will look into the security threats
in MEC then dos attacks and defense
mechanisms then we will look in into the
different type of attacks in the edge
Computing system such as flood based
attacks zero day dos attacks sign side
Channel attacks and defenses data
perturbation various type of injection
attacks authentication and authorization
attacks access control and
Authentication Protocol so we will
confine these these kind of uh attacks
or security threats in the edge
Computing scenario
only so let us see about the Mec
Paradigm here you can see that in this
particular figure that here there exist
a basis station and along with the basis
station you will see this kind of MEC
servers are there so you can see that uh
the deployment uh so what is being shown
over here in the com conventional cloud
computing so conventional cloud
computing does not have this Edge
whereas this is The Edge Computing so
along with the base station you will
find the Ed servers so after the edge
server you will get the cloud so this is
the
cloud this is The
Edge so Edge servers are with the Bas
station
and here are the devices so the devices
or the
applications which are latency sensitive
applications they have to now being
served near to the basis station or near
to the source they are called
MEC multi access Edge Computing so as
the mobile as the user or the
application is mobile then it may
reconnect to some other kind of uh
MEC and if let us say beyond that if it
is not a latency sensitive application
then it may offload the computation to
the
cloud so when the load comes into the
system of this kind of 5G or 6G 5G then
it has to be identified whether it's a
latency sensitive
application or or something else then in
that case if it is a latency sensitive
application whether it requires multi
access Edge Computing or a cloud
computing that has to be now decided
with the due intelligence and here this
is the scenario when Edge Computing was
not there was only the cloud computing
so the computation offloading or task
offloading has to be done
with the latency to the
cloud and these cloud data centers are
located at different part of the globe
so therefore cloud computing are having
the drawback such as it is having a huge
amount of latency Jeter locational
unawareness insecure data Outsourcing
and limited access and bottle legs now
with this they introduced is MEC so MEC
is in produced now uh in between the
cloud and the basis station called
multiaxis Edge Computing the benefits
which this particular multiaxis Edge
Computing will bring into is ultra low
latency so the latency now becomes an
ultra low latency so most of the latency
sensitive applications are now possible
in this particular MEC
Paradigm the second important
aspect is called locational unawareness
so this is to be changed to location
awareness why because now the Computing
is done in that vicinity of that user so
therefore all the location information
is being preserved unlike here the all
the data if it is being sent to the
cloud that means location awareness it's
a location unaware type of computational
offloading is done now with the Mec this
is the location awareness so this is
very very important so many type of
location based services are not possible
in this multi access educ
Computing then next important thing is
in proximity data
Outsourcing so in proximity data
Outsourcing
gives that it removes this insecure data
Outsourcing with the inom in proximity
data Outsourcing now it is having the
improved capacity at the edge which was
not there earlier now it has the higher
bandwidth why because most of the uh
most of the computation is done very
close by so back hall bandwidth is
already available so hard bandwidth is
available and realtime responsiveness is
there so with this now you know that the
applications such as video streaming AR
and VR tectile internet MTC M2M and V to
e there are different type of
applications which are now possible in
this particular uh Edge mobile or multia
Edge Computing Paradigm now let us see
about the security threats uh in this
particular domain that is called Edge
Computing so the edge Computing is a
paradigm that shift the data processing
services to the network Edge that means
where it is shifting from
cloud so the edge architecture provides
a faster processing and response among
various other benefits so emergence of
Ed Computing introduces more security
threats since it
increases the real world
attack surface from the following four
angles so let us see why the security
threats are more here when you talk
about the edge Computing first is that
weak computation power now you know that
compared to the cloud servers the
computation power of the Ed server is
relatively weaker so this weaker form of
computing power with the edge is one of
the weak computing power is is is
attracting towards the security threads
so the head server is more
vulnerable to the existing attack and no
longer be effective against the cloud
servers so Cloud servers are more secure
compared to the Ed servers now compared
to the general purpose computer Edge
devices have more fragile uh defense
system and as a consequence many attacks
may have been ineffective against the
desktop can impose serious threats to
the edge devices so second security
threats in Edge Computing is called
location is called attack
unawareness so unlike general purpose
computers majority iot devices do not
have the user interface regardless of
the fact that some may be having the LED
screens so use user may have the limited
knowledge about running status of the
device and even if an attack is taking
place in an edge device most users will
not be able to so therefore attack
unawareness is another issue or a
security threat here in Edge devices or
Edge Computing now coming third threat
is called operating system and the
protocol heterogeneity unlike general
purpose computer that tend to use
standardized operating systems and
communication protocols most of the S
devices have different operating systems
and protocols without a standardized
regulation this is also and this problem
uh directly uh LEDs to diff the
difficulties of Designing a unified
protecting protective environment for
the edge
Computing finally it's a course grain
Access Control uh system which is being
implemented in Edge Computing so it is
Access Control models are designed for
general purpose computer and cloud
computing mainly consist of four type of
permissions no read write read only
write only read and write so such a
model would never be satisfiable in a
edge Computing due to the more comp
complicated systems and their enabling
applications which call for fine green
access control that should handle
questions such as who can access this
sensor by doing what at when and how
unfortunately current Access Control
models are more are mostly course
grained so let us go inside with this
understanding The Edge Computing
security and privacy so attacks
targeting The Edge Computing
infrastructure have drastic rise in the
recent
years one of the most notable attack
happened in the real world is Mirai
attack which Mirai virus it managed to
compromise more than 655,000 iot devices
within first 20 hours after its release
in
2016 by exploiting the devices weak
authentication
vulnerabilities a new days a few days
later these compromis deves were turned
into the bot Nets to launch Doos attack
against the Ed server shutting
down around 178,000 domains shortly
after that variations of Mirai such as
iot Reaper and
hazami were captured and they were
believed to infect more than 378 million
iot devices in
2016 since the discovery of the first
Mirai botet in
2016 iot botet attack were reported to
cause the damages worth 100 billion US
by September 2018 these numbers only
indicate the attacks and the asset
losses that were officially identified
and reported the total amount of
undirected attacks and losses should be
much more
higher so if you see the Ed Computing
security and the
Privacy uh that is the percentages
account
according to the uh type of attacks T
targeting The Edge Computing
infrastructure that has happened in the
real uh world so what you will find that
uh the most prominent of these attacks
are DOs attacks and malware injection
attacks with a
26% of its
share that is
dos and the malare
injection whereas side Channel
attack contributes to 8% of the edge
Computing security and privacy bad data
induction attacks is
9% man in the
middle is is a 9% and authentication and
authoriz attacks is because of
22% so let us see that this
is these are the most important security
attacks that we are going to see in the
edge Computing scenario so we will now
discuss summarize the
major state of the art security threats
and the attacks faced by The Edge
Computing we then explore the
corresponding uh defense mechanisms
which are either the detection based on
based for the purpose of organizing the
attacks or prevention based
for objective of obstructing the attacks
from happening lastly the outline we
outline the root causes of the attacks
and discuss the Practical practicality
of launching and defending them so let
us start with the DS attacks and the
defense mechanisms so D do refers to the
type of Cyber attack in which the
attackers aim to disrupt the nor
Services provided by one or more
servers based on the distributed
resources such as the cluster of
compromised Edge devices such as they
are called also known as bot
Nets now this
particular attack is a powerful attack
that aims to prevent the legitimate use
of
service a traditional dos attack occurs
when an attacker persistently sends a
stream of packets
to a victim from compromised
distributed electronic
device thus the hardware resources of
the victim are quickly exhausted for
handling malicious packets and can no
longer be the
process any legitimate request on the
time in some other denial of service D
do attack scenarios an attacker
persistently sends
malformed p
that confuse an application or a
protocol of the victim to falsely
conclude that all the channels and the
resources are
occupied now compared to the cloud
servers S servers are more susceptible
to the Dos attack since they are
relatively computationally less powerful
to maintain a strong defense system as
the cloud servers can defend Ed servers
mainly provide the
services to the edge devices that are
well known to the error prone in the
regard to the security settings due to
the computation limited hardware and
heterogeneous firware having noticed the
error prone SED devices attackers favor
first compromising a number of s devices
turning them into the weapon against the
S servers so here you can see that this
attacker will take a control of these
particular
devices and they are now become the
Rogue Edge devices and this is the
typical architecture of a DS attack and
then use this particular devices to
inject
unnecessary or information into the
system
thereby avoiding this particular s
server uh to respond with all that uh
which legitimate kind of uh requests so
flooding based attacks uh is quite
prevalent uh and a type of DS attacks
aiming to shut down the normal service
of the server based on the large amount
of flooded malun malformed and malicious
Network packets are mainly classified
into UDP flooding icmp flooding sink
flooding ping of death sttp flooding and
slow risk according to the uh attack
techniques so in a UDP flooding attacks
attack continuously sends large amount
of noisy UDP packet to the Target
servers causing the server incapable of
handling the benine UDP packets in time
thus interupting the normal UDP Services
provided by the UDP The Edge servers
icmp uh flooding attack similar to that
particular UDP flooding attack the
attacker exploit the icmp protocol to
craft an attack by sending large number
of icmp request packet to the Target
server as fast as possible without
beating for the
response now zero day attacks is more
advanced than flooding based dos attacks
earlier but it is more difficult to
implement the zero day dos
attack and here an attacker finds an
unknown vulnerability that is zero day
vulnerability is a piece of the code
running on the target Edge server or a
device with can cause the memory
corruption and finally result into the
service
shutdown now current defense solution
the root cause of flooding based attacks
is the protocol level design flaws and
vulnerabilities within the network
communication protocol while the root
cause of zero day attack lies in the
code level vulnerability that can
trigger the memory failures and the
Corruptions so the current defense
solution against the flooding based
attack mainly adopt to the to detect the
Fielder philosophy
while those against the zero attack uh
zero day attacks mainly focus on the
code level one ability
identifications now detection of the
flood based dos attack can mainly be
classified into two categories per
packet
detection the other is statistics based
detection per packet detections aim to
detect the flooding based attacks at the
packet level since the flooding based
attack is launched mainly by by sending
enormous amount of malicious or
malformed network packet detecting and
filtering those packets can have an
effective defense when a suspicious
packet is identified the network can
simply drop the packet before it arrives
at the destination uh at servers
although this may sound trivial
detecting whether the packet is a Dos or
a oriented malicious one is never
see now to defend again zero day attack
the researcher have developed the
mechanism such as pointer tainted
detection and ECC memory to spot
possible memory leaks in a program such
methods require presence of the original
Source codes which are usually
unavailable for the S
devices so there are several researchers
propose the approach to perform the
memory analysis based solely on the
firware so there are various um deep
learning models also recurrent neural
networks graph neural networks and deep
natural language processing can identify
these vulnerabilities in the firware
with higher accuracy rates now coming to
next type of attack which is called side
Channel attack and let us see its
defenses so side Channel attack refer to
those that compromise the user security
and privacy using any publicly
accessible
information uh that is not privacy
sensitive in the nature namely the site
Channel information such public
information is typically correlated
secretly with certain privacy sensitive
data that should be protected attackers
then explore the hidden correlations to
finally infer the protected data from
the side channels since any public
information can have the potential to
link some of the sensitive data site
Channel attack can
happen anywhere in the edge Computing
architecture an
attacker constantly obtain the side
Channel information from the target Edge
Computing infrastructure then fits it
into the specific algorithms machine
learning model that output the desired
sensitive information the most popular
site channels in the edge Computing
include communication
signals Electric power consumption
smartphone or embed sensors the attacks
exploiting the communication Channel
happens when the attacker continuously
monitors the transmission between two
Edge nodes those exploiting the power
consumption
often occur after after the attacker
steals the power consumption data of of
the Ed devices and those exploiting
smartphone based channels happens when
the attacker secretly accesses the
smartphone and is SE the information is
stored in publicly available oblique uh
procedure files are generated by the
embedded sensors so here you can
understand this particular attacker so
it can obtain using the available
information that is traffic streams
power consumption timing Etc then it
will use this information to infer the
private information location password
activities and so on that will iner and
then it will Target The Edge computing
system and this is the typical
architecture of a site Channel attack so
attacks which are exploiting the
communication channels in the edge
Computing exploiting the communication
signals has a high potential to reveal
the sensitive information of a victim
due to the rich Channel information so
attacker may can be any curious
malicious node which does not have an
edge device or an ed server who
constantly who continuously sniffs the
network traces and wishes to extract
the sensitive information out of that
this type of attack can further be txom
ised into two sub classes those
exploiting the packet streams and those
exploiting the wave signals so power
consumption is an power consumption is
an indicator of an electric usage of a
system it carries the information
related to either the device that
consumes the energy as the different
device have different power consumption
profiles when operating or intensity of
or intensity of the computation in the
Computing task it attracts the attention
from number of researchers to
investigate its link to the sensitive
data this type of attack can be
classified into two subclasses attack
exploiting the power consumption
collected by the meters and those
exploiting the power consumption
collected by the oscilloscopes so
smartphones are the key Edge devices in
many applications different from iot
devices smartphone have advanced
operating systems and a rich richer
system information compared with the
less Advanced iot devices smartphones
can be exposed to a broader surface
attacks so attacks can be classified
into two subclasses attacks exploiting
the SL proc file system and those
exploiting the smartphone embedded
sensors so current defense solution so
so the root cause of the side Channel
attacks is hidden correlation which
which could be very complicated and
hardly
identified between the sensitive data to
be protected and publicly available side
Channel now defenses against the side
Channel attack can be performed from the
two directions restricting the access to
the site Channel information and
protecting the sensitive data from the
inference attacks so there exist no
fible defense mechanisms that can
restrict the access to the uncontrol Lev
channels leaving the sensitive data
protection the only approach for such
kind of
attacks data
perturbation data
perturbation most well-known data
perturbation algorithm to protect the
sensitive data from inference attacks is
K anonymity which modifies the
identifier information of a piece of the
data before publishing its sense
attributes making it indistinguishable
from another K minus one pieces of the
data with these K pieces of the data
forming an equal class so K anonymity
suffers from homogeneity attack when the
values of a sensitive attribute within
the equivalence class are identified to
overcome this
issue IE diversity is is being proposed
by ensuring equivalence class to have at
least I distinct values for each
sensitive attribute so restricting
access to the side Channel attack an
alternative approach to defend against
the side Channel attack is to restrict
the access to the side Channel
information side Channel offers cation
on the source code level is such a
scheme to defend from software
surface so moldar proposed the mechanism
to eliminate the control Flow side
attacks from the C code and there are
another walk which has developed the
site Channel detection scheme to monitor
the abnormal cache behaviors on cloud
and Ed
servers these two
methods directly part up the site
channels to obstruct the accuracy of
inference
algorithms now recent years the
development of trust June technology
advances
quickly and this method mainly allows
unauthorized access to the side channel
that is being protected in the trust
Jone so malware injection attack is
shown over here so here there exist an S
device and here there are S servers now
attacker sits in between now it will
inject the malware and it will inject
the malware to the S devices a typical
architecture here the malware injection
attacks so the action to effectively and
stealthily inject install malware into
the Computing system is called malware
injection attacks this type of attack is
one of the most dangerous ones since
malware is a significant threat to the
system security and data Integrity so
traditional internet or a general
purpose computer infrastructures where
strong computational power is available
to support high performance firewall or
other threat protection system malware
injection is not always feasible and
possible there now typical architecture
for malware injection attack is
straightforward the objective of the
attack is to inject malware malware
quots to the S devices or the servers so
malware injection attack in the edge
Computing can be classified into two
categories the server side injection
that is injection attack targeting the
servers S servers and device side
injection that is injection attacks
targeting to the edge
devices now server side injections there
are four different type of injection
attacks targeting the S servers namely
the SQL injection xss csrf ssrf XML
signature wrapping now SQL injection is
the code injection techniques that
destroys the backend
servers to construct a normal SQL query
the legitimate user is allowed to
manipulate only the the designated areas
to get the result from the server
however attacker May manage to
circumvent this constraint by inputting
Escape characters along with the query
string
now cross site scripting accesses is a
client side attack in which an attacker
injects malicious codes into the data
content which can be accessed and
executed automatically by the server
note that the client side here does not
refer to the edge device side but rather
The Edge server site in which an edge
server works as the client to visit or
access the service is provided by other
Edge servers or the cloud
servers therefore contrary to the
traditional general purpose Computing
system accesss is a type of injection
attacks that happen to the S server
level the server Computing the attack is
caused by the fact that the Ed servers
do not filter the code from the data
content xss is not a novel attack and
its mechanism has been studied it is a
serious threat to the Ed Computing
infrastructure now device side
injections now various diverse methods
for injecting malware into the iot
device exist since iot devices are
highly heterogeneous both in the
hardware and the firware most commonly
approach
to remotely inject malware to exploit
zero day vulnerabilities that can lead
to remote code
execution one of the most infamous
example
of of iot Reaper virus captured in 2017
that infects millions of iot devices and
Wi-Fi by exploiting at least 30 RC
vulnerabilities in nine different iot
ranging from Network router to IP camera
now
Q QE
discovered uh RP rfu that is remote
firware update protocol by adopting the
laser jet printer allows an attacker to
modify a pre deployed firware or the
printer due to the lack of signature
verification
checks similarly you know that all these
are different type of now defenses
against the server side
injection uh cons uh also consider four
major type of attack SQL injection
excesses csrf oblique ssrf and XML
signature wrapping so SQL injection
attacks were discovered at the time when
SQL databases were
invented defense and detection
mechanisms have been studied for Fairly
long
time now detection
based detection Focus techniques
basically emplo the code checking with
various schemes such as static analysis
Dynamic debugging blackbox testing taint
based analysis
while prevention focused attack
techniques Target to prevent any illegal
SQL queries from being executed by means
such as setting up a proxy filter
employing instruction set
randomization now defense against the
device side injections so injection
attacks targeting iot devices the main
threat comes from the firware
modification attacks ly limited research
exist to investigate the corresponding
defense
mechanisms
now uh there are various ideas are
floated based on address space layout
randomization instruction set
randomization and so on and various
papers or researchers they have also
proposed the
software symbiotic method that injects
which injects intrusion detection
functionality into the boundary firware
of the existing iot
devices so the two mechanisms out of
that are quite reasonable their
realizations are now easy and therefore
specific methods on implementation of
the two mechanisms are being looked upon
so here you can see in this particular
figure that this authentication and
authorization systems and here you have
the target Edge computer system whereas
the attacker is
concerned if the direct access is
blocked then unauthorized access is
bypassed and you know that it
can make an attack so authentication is
an action of verifying the user
identities who request certain services
so authorization is a
process determining the access rights
and privileges of an entity confirming
that entity behaves based on the rights
without crossing the
boundaries
so authorization is usually proceeded
with with the authentication for
identity verification so in Edge
Computing authentication is generally
performed between the S devices and S
servers under certain circumstances it
also performed that the S devices or
among the S servers in a decentralized
way is also
called Trigger action platforms so
authorization in the edge Computing
often refers to the activity when an ed
server grants permission to a certain
device or
application yet it is also possible for
the devices or the application to Grant
permission to the other devices in a
Trigger action scenario that is in the
home
automation so attack specification of
authentication and authorization attacks
if an attacker intends to directly
access the protected Edge servers or the
devices it would be blocked by
authentication system that we have shown
in the previous diagram therefore the
attacker seeks the method to bypass the
authentication process performing
unauthorized access so attackers can be
classified in four types dictionary
attacks attacks exploiting the
vulnerabilities in the Authentication
Protocol attacks exploiting
vulnerability vulnerabilities in in
authorization protocols and
overprivileged attacks with the first
two targeting authentication protocols
and rest are tar authorization protocols
now coming to more details about the
diary attacks so authentication based
attacks POS significant threat for
securing The
Edge Computing so the most
straightforward authentication based
attack is the dictionary attack in which
an attacker processes a dictionary
containing the most mostly used
credentials password and the inputs of
all possible credentials password in
this dictionary
now this type of attack is also known as
Brute Force attack in terms that is
widely used in the industry now
exploiting weaknesses in the
Authentication Protocol so diary attack
is an easy to launch on the other side
it has a serious drawbacks such as high
resource consumption and low success and
a low success rate
therefore it is
investigate researchers are
investigating more attack by discovering
the design flaws of the authentication
protocols there are
various works or various proposals they
have been uh mentioning over here which
are still under the uh development that
is WPA TLS based authentication and
there are many other which are detected
here exploiting the weaknesses in the
author authorization protocols is
another way of this authorization based
attacks usually exploit the weaknesses
or the design flaws existing in
authorization protocols in order to
achieve the unauthorized access
sensitive information or perform
privileged operations also known as
overprivileged issues so Edge Computing
o author o
authorization uh is a widely used
authorization protocol designed for
multiparty
authorization so uh these are all uh
being uh now under
that so overprivileged attacks uh you
know that basic problem of this uh
Authentication Protocol is that
overprivileged issues uh in typical
authorization systems now current
defense solution for authentication and
authorization attacks the root cause of
the Diary attacks is the adoption of the
we credentials in the in the
authentication while the root cause the
other three type of attacks is the
protocol level design flaws or
implementation level design flaws so
this authentication is the security
entry point of a system and most
attacker take down the edge devices or
sver beginning with the compromising the
authentication systems at the first
place therefore securing authentication
and Edge is quite
crucial so there is a fair amount of
work exploiting the diary attacks to
break possible POS weak credentials in
the edge
Computing first is due to the limited
computational power issues in the edge
infrastructure second is unlike
traditional internet applications Edge
Computing infrastructure have have we
more
subscribers that is called Edge devices
third is storing credentials in the iot
is not secure so these are some of now
hardening the Authentication Protocol is
to defense against the attacks
exploiting the
vulnerabilities is to choose to enhance
the security of the communication
protocol or the secure cryptographic
mechanisms that is what is uh being uh
taken into into the way of by
different uh
researchers now hardening the
authorization protocol so authorization
in h Computing infrastructure is equally
important compared to the authentication
system systems without properly
designing the authorization mechanisms
an attacker can exploit the weaknesses
in the authorization systems as
well so there are many implementation of
O authentication protocols and different
versions are even theoretically secure
then there are various works and o o
authentication service provider such as
Google Facebook and Cena they they first
check and then a lot of new ways are
coming which can be exploited using
apis now defending against the or
privilege attacks or privilege issues
mainly exist in iot based systems and
mobile
systems so there are various scientists
first they propos the NLP based method
to check the inconsistencies between iot
application implementation and and its
description in order to identify the
overprivileged
exploits there are different
researchers they applied different
methods such as Tain based analysis
Information Tracking technique and then
uh identify 20 plus 20 smartphones using
iot guard and these mechanisms need
authorizers to be able to access the
source code of an iot which may not be
available at the
time now comparison study on the
security issues in the edge comp
Computing and the cloud computing so
cloud is a practice of utilizing the
centralized remote servers and the data
centers in the internet for storage
processing and Analysis the end devices
in the cloud are
fully far fleshed computer system
connect to the cloud
servers note that
cloud system such as smart
things and units and iot devices usually
delegate most of the services to the
local hubs can handle iot as they not
strictly the cloud such systems are now
Edge Computing system Ed on other adopts
decentralized hierarchial design that
may involve low profile devices and
computers as as
servers the end units is an edge
Computing system are typically the iot
and the mobile and they are much more
resource constraint compared to the
forged
computers and they are connected with
Wi-Fi and 4G so Doos attacks in the
flooding based Doo attacks attacker
tends to exploit the vulnerability of
the network protocol of the target
system so cloud computing typically
employs the heavybit network protocols
such as sttp htps FTP SMTP Cloud servers
are more computational
resources for strong security protocols
while the edge Computing typically
adopts lightweight Network protocols
such as coap and qtt UDP due to the
resource
constraints so such a difference makes
the attacking effort significantly
larger in the cloud computing System
since the compromising the far- fledged
computers becoming more and more
difficult using the firewalls running on
them so DS attack May no longer
effective in in in cloud computing but
so side channels attacks if you if you
understand the attackers have the wide
attack surface to launch side Channel
attacks in the edge Computing as the
edge devices are typically connected
wirelessly and are more approachable so
attackers in the cloud often choose to
extract valuable
information and cloud computing is
typically vulnerable to the side Channel
attacks that exploit the packet based
communication in Edge Computing
attackers have more side channel to
manipulate and that can easily access
the divers side Channel
information now malware injection
attacks now attacks targeting the cloud
computing attackers tend to First launch
the server side injection to take down
the cloud servers and then infect the C
devices that
compromise The Edge the cloud servers
while the edge Computing attacker tend
to First launch the device side
injections to take down the number of s
devices and then abuse them to the S
servers so the reason behind such a
phenomena lies in the client devices in
the cloud are usually do
interconnected and taking down few of
them may have a limited impact on the
other hand s devices in the edge
Computing are typically interconnected
and thus taking down a few can have much
more significant impact if the attacker
spreads his attacks among the
interconnected devices such compromising
Cloud device is much more difficult than
compromising uh The Edge device malware
injection attacks are more specific and
popular in the edge Computing systems
authentication and authorization attacks
both the edge and Cloud are vulnerable
to the diary attacks Edge devices have
higher chance of implying weak
credential than those cloud computing
making the diary attacks more popular
and practical in Edge Computing so Edge
Computing vulnerable two attacks
exploiting the weaknesses in the
wireless based authentication protocols
such as WPA and WPA2 Bluetooth 4G 5G
while cloud computing is vulnerable to
the attacks exploiting the weaknesses in
those SSL and TLS that can be applied to
the Wi Network on the other hand even
though both Ed and Cloud are vulnerable
to the attacks exploiting the weaknesses
in the authorization protocols Edge
Computing is more
susceptible compared with the cloud
because cloud has relatively simple
authorization scenarios due to the
centralized design while involving less
number of communicate communication
parties at servers are deployed in a
decentralized manner involving more
communication parties complicating the
whole authorization process in addition
over PR attacks mainly occur iot and
mobile devices they are typically
specific to the edge Computing System
since such systems have much more
complex Access Control scenarios and and
general purpose
Computing so the root causes of security
issues protocol level design flaws are
the fundamental support for the basic
functions such as communication data
processing The Edge Computing from the
therefore designing secure protocol
critical to maintain healthy Edge
Computing ecology unfortunately many
current protocols adopted by have the
design pla so designer have to focus on
utility and the user experience treating
security as unimportant if if
unnecessary so by exploiting this design
flaws attacker have
achieved can achieve the attack goals
ranging from Simply shut down a normal
service DS to fully controlling an edge
device or the server that is malware
injection tax so implementation level
flaws implementation is the process of
practically realizing the edge Computing
functionalities so protocol may be
proven secure mathematically does not
necessarily mean that implementation is
secure so logical flaws in
implementation can neutralize the
security strength of a protocol so there
are two main reasons leading to the
implementation flaw developers May
misunderstood the foundation of the
protocol migrating a protocol from other
platforms to the S server may cause
inconsistency so implementation flaws
can equally serious as the protocol
level design Flaws by exploiting the
implementation flaws and and attacker
can bypass the security feature provided
by the
protocol now code level
vulnerabilities code is the basic unit
of the program that defines the exact
execution as mentioned many attacks that
is the code level vulnerability is being
introduced the implementation level flaw
code level vulnerability are two
different concept with for with former
mainly referring to the logical flaws in
the Practical
realization so the presence of such
vulnerability lies thoughtless
programming process when quotes for the
millions lines of the quotes
vulnerabilities an attacker can achieve
the attack goals ranging from Simply
shut down a normal service to fully
controlling n servers in malare
injection attacks
now data correlation that is in Edge
computer system tons of data would be
generating from The Edge infrastructure
constantly with some of them sensitive
placed under the strict protection while
the others are less sensitive and
exposed to the public without now there
exist hidden correlation between
sensitive and insensitive data which may
not be
straightforward nevertheless exploiting
this correlation and leveraging various
attack models can infer the sensitive
data that is side Channel attack or even
tamper them based on the insensitive
data lacking the fine grain access
control is one of the most important
security measures access control and
through which an entity is permitted to
access only at least the the resources
it needs so such a measure has a good
reputation in protecting the traditional
general purpose system for a long time
but it cannot be directly adopted in the
edge Computing due to the more complex
and fine gr scenarios now many Edge
current Edge Computing system Implement
only the course grain Access Control
mechanisms and even do not imply any
access control mechanism lacking the
fine grin access control is specifically
designed for Edge computer application
May significantly lower the bar of
launching an
attack that is man in the middle so
conclusion in this lecture we discussed
various type of security threats in in
the Mec that is multi access Ed computer
shoting and then we talked about
different security threats and the
defense mechanisms in the edge Computing
system we have also covered about how
the fine grain and course grain Access
Control are the issues in the edge
Computing and the authentication
protocols thank
[Music]
you
[Music]
[Music]
[Music]
