IEEE Consumer Electronics Magazine - January/February 2024 - 99

services are outsourced to support operations
at the network's edge.7
To address the abovementioned challenges,
in this article, we briefly examine why the integration
of edge computing and blockchain technology
is essential for next-generation IoT
applications and how blockchain will be an ideal
choice to maintain privacy, security, and trust in
the data.
SECURITY AND PRIVACY ASPECTS IN
INTELLIGENT EDGE COMPUTING
Edge-centric intelligent applications depend
on information gathered from a variety of sources.
As a result, the effectiveness of these
applications is determined by the data. Enemies
can take advantage of this critical reliance
by manipulating or poisoning data in
order to affect intelligence systems. Because AI
systems are meant to predict, categorize, or
extrapolate from acquired data. An adversary
might use several methods to influence these
functions, such as data deception, model poisoning,
or a privacy attack.6 Figure 2 represents
a typical distributed intelligent edgecomputing
scenario where a resource-rich
edge server combines data from heterogeneous
edge devices. This information is
yielded with numerous edge servers for essential
data analytics via model training. Likewise,
several venerability issues may arise at varying
levels of intelligent edge networks.
Manipulating Training Sample
An attacker can manipulate a learning process
by directing the targeted devices and providing fictitious/manipulated
data to damage the training
models, also known as data poisoning. Data
poisoning would not only result in inadequate
local training, but it would also contaminate the
functionality of programs, causing them to make
incorrect decisions. Consider autonom-ous driving,
when data poisoning attack froma specific IoT
device (e.g., a camera) may cause the car to drive
left due to a faulty learning result, whereas driving
right is the proper option. However, protecting
against such poisoning attacks is difficult due to
the limited capabilities of IoT devices while supporting
complicated security algorithms.
January/February 2024
Extrapolate Private Data
Data must be transmitted from low-level devices
to edge servers and even to the faraway public
cloud in intelligent edge computing. In
addition, cloud servers, edge servers, and endusers
all demand user information and various
sorts of message exchanges. In this situation,
attackers may hand off the message sent over
the communication interface and deduce secret
messages. Another difficult challenge in intelligent
edge computing is how to defend personal
information sharing for small devices.
Training Agents Attack
Compared to traditional cloud technology,
learning agents in edge computing are dispersed
over the edge layer. As a result, they are far more
vulnerable to assault than cloud agents. For example,
reinforcement-learning applications require
agents to interact with their surroundings in order
to optimize their rewards. A skillfully timed strike
and a luring attack can readily target an agent in
these circumstances. When an agent is assaulted,
it affects the entire edge-learning process. For
example, the attackermight combine a concealing
input vector into one learning agent. When learning
agents interlock with one another, all learning
agents inside the edge networks get affected by
that strike vector.
COLLABORATIVE EDGE-COMPUTING
FRAMEWORK
Edge computing reduces latency, energy, processing
times, and communication overhead across
all devices while delivering a high quality of cloud
serviceclosertothe networkedge.8 However, providing
adequate security in a dispersed edge environment
is challenging. Because data processing
occurs outside the network, identity theft and
cybersecurity breaches are a risk. Blockchain is a
solution for storing data that makes it difficult or
impracticable to alter ormodify.
Blockchain for Communication
Computers attached to the blockchain system
obtain a copy of the blockchain from edge devices,
which are the structure bars of the blockchain
network. Instead of being updated by a
central computer, these edge nodes communicate,
store, and update information over the
99
IEEE Consumer Electronics Magazine - January/February 2024

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - January/February 2024
