IEEE Consumer Electronics Magazine - September/October 2021 - 15

process all the data from the sensors to navigate
through the traffic effectively and efficiently. In
order to do so, we delegate the computing task
to external edge nodes, and the architecture is
called vehicular edge computing (VEC). Since we
are offloading the processing to an edge node,
the edge node should be secure and smart as
well. This can be achieved through the use of
blockchain and artificial intelligence (AI) in VEC.
Blockchain and AI can guarantee smart decision-making
capabilities, data security, and network
security in a VEC system.1-3
Despite such potentials, the blockchain-based
VEC systems encounter several issues such
as spam attacks, targeted distributed denial of
service (DDoS) attacks, malicious contracts, etc.4
In order to nullify these types of attacks, the
affected peer nodes can be removed from the overall
consensus process. The process of removing
mining peer nodes from the consensus is called
miner node selection (MNS). The selection and
classification of theminer node on the basis of reputation
will ensure that only infected nodes are
removed from the consensus process, which can
be achieved with the help of AI. The Blockchain
provides immutability and data security5 while AI
provides autonomy and greater decision-making
capabilities such as MNS, anomaly detection, etc.,
in the field of blockchain.
The application of blockchain in a VEC system
with AI-enabled MNS can collectively solve the
issues in the VEC while at the same time tackle the
issues within the blockchain-based system. This
type of comprehensive architecture has not been
done before. In our implementation as shown in
Figure 1, we have chosen hyperledger fabric (HF)
for implementation, which uses the practical byzantine
fault tolerance (PBFT) consensus mechanism.
The PBFT consensus mechanism requires
two-thirds majority of nodes in the consensus process
to add a transaction to the blockchain. If the
two-thirds majority could not be obtained during
the consensus process due tomalicious nodes, the
transactions will not be added to the blockchain. If
the edge nodes, which are RSUs in our case, are
removed from the consensus process with the
help of MNS, the required majority will be reached
ensuring the smooth performance of the system.
For this purpose, we discuss the mechanism for
RbMNS in the blockchain-based vehicular edge
computing (BloVEC) system. The implementation
is achieved through the artificial neural network
September/October 2021
(ANN) for the reputation model and HF for MNS.
The main job of MNS is to select nodes that participate
in the consensus process. Additionally, in our
system, we have used telemetry data consisting of
GPS coordinates, speed, acceleration, etc., as transactional
input. The telemetry data are used for
tracking and analysis of the traffic for navigational
purposes. The data are collected through various
in-vehicle sensors. With the application in navigational
purpose, we can discuss various routing protocols6
that can be used such as topology-based
routing, position-based routing, cluster-based routing,
etc. However, this article solely focuses on
MNS in a BloVEC system.
MOTIVATION AND CONTRIBUTION
BloVEC has a very high potential to offload
computation from the connected vehicles and
at the same time make the data and system
robust and secure. However, a compromised
miner node can degrade the performance of
the system or halt the addition of transactions
in the system. In order to overcome this shortcoming,
we needed an architecture that
detects and removes the compromised miner
node from the consensus process in BloVEC.
To realize this, we chose the HF platform,
which uses PBFT7
as itsconsensusprotocol.
Similarly, there are many consensus mechanisms
other than PBFT such as SBFT,8 TPBFT,9
etc. The selection of the best consensus
mechanism is a research avenue on its own.
Since we are using HF, which comes default
with the PBFT consensus mechanism, we are
using the PBFT consensus mechanism in this
article. We know that these BFT algorithms
require 3f þ 1 nodes to tolerate Byzantine fault
up to only f nodes.7 If these faulty nodes are
not detected and removed, they may break
system security. Despite this fact, PBFT lacks a
mechanism to detect and remove compromised
nodes. In the VEC application, the
vehicles send transactions to the RSUs, which
endorses the transaction and adds it to the
blockchain. If the RSUs are compromised, they
will endorse the transaction with the anomalous
value making it harder to reach the twothirds
majority to add the transaction to the
blockchain. Therefore, to make the blockchainbased
VEC system more secure, we have to let
only good nodes take part in the consensus
mechanism.
15
IEEE Consumer Electronics Magazine - September/October 2021

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - September/October 2021
