IEEE Consumer Electronics Magazine - January/February 2024 - 23

The possibility of tampering the system logs is a
matter of concern in such an approach.4
In this article, we revisited the research chalFIGURE
1. Hierarchical representation of edge,
fog, and cloud-computing environments.
(2021-2026).b The devices at the edge of the network,
also referred to as edge devices, are not
only collecting data or sensing the surrounding
environment, but also preprocessing the data and
learning the basic meaning of those data. This is
done by implementing an essential intelligence
atop the limited computing resources available
with the edge devices, also known as edge intelligence.1
In such a scenario, the edge devices are
primarily managed by the central control center
that usually resides in a cloud-computing environment
andworks in a master-worker approach. Furthermore,
to improve the quality of service (QoS)
through several supplements, such as lesser
latency and response time, fog computing environment
is introduced between cloud and edge
computing layers,2 as shown in Figure 1.
However, in the case of clustered edge intelligence
(CEI), the main focus is on creating context-based
intelligence clusters. In doing so, the
primary focus has now shifted from clustering
of devices to clustering of intelligence of several
edge devices, irrespective of their location
and other characteristics. Such an approach
may help reduce the system complexity.3 While
doing so, it is highly essential to consider the
attributes of underlined edge devices, such as
mobility nature, the type and purpose of the
devices, what kind of data the device generates,
and so on.
One of the research challenges in CEI is the
ability to track the chain of tampered-proof historical
events, such as failure of actions/devices and
performance degradation.4 One of the conventional
ways to track a past event is to analyze large
size system logs collected from all the devices.4,5
b
[Online]. Available: https://www.mordorintelligence.com/industry-reports/
internet-of-things-moving-towards-a-smarter-tomorrow-market-industry
lenge as mentioned previously. We proposed a
blockchain-based security mechanism that would
enable the administrators/users to track the chain
of immutable events to analyze and investigate
the reason for any anomaly event. Blockchain is
one of the most suitable candidate technology for
the implementation of several distributed systems.3,6
This would secure the details of the
device's actions/events and establish the chain
among themwithin the clusters. Due to its promising
features, such as distributed, traceability of
shared information, transparency, immutability,
and reduced costs, the application of blockchain
technology is diverse, ranging from edge intelligence7
to vehicular network8 to different areas of
smart city.9 There are several works that used
blockchain for edge computing and they are
summarized using Table 1. Majority of these
approaches do not follow cluster or distributed
approach, and they are also expensive in terms of
complexity and resource requirements.
BACKGROUND
This section gives a brief overview of CEI and
how it differs from generic edge intelligence.
Upon realizing the cloud-computing capabilities
in traditional methods, most intelligence is
imposed on cloud servers following a centralized
system. However, to mitigate the limitation of
cloud computing itself, research has started to
bring the intelligence to the proximity of devices
at the edge of the network, referring to as edge
intelligence.1 In some cases, edge intelligence is
also referred to as deploying a minimal version of
artificial intelligence (AI) on resource constraint
edge devices.
To further minimize the network burden on
the energy constraint edge device and reduce the
systemcomplexity, the edge devices are clustered
based on their type, location, or specific feature.
Hence, the driving factor to cluster the edge devices
is tomanage the devices and utilize the limited
resources in an effective manner. For this, most of
the research works focused primarily on the
device-level management rather than the intelligence
intended to be implemented in devices. The
device-centric clustering mechanism introduces
January/February 2024
23
https://www.mordorintelligence.com/industry-reports/internet-of-things-moving-towards-a-smarter-tomorrow-market-industry https://www.mordorintelligence.com/industry-reports/internet-of-things-moving-towards-a-smarter-tomorrow-market-industry
IEEE Consumer Electronics Magazine - January/February 2024

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