Page 1

1

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

ISSN: 2178-9010

DOI:

http://doi.org/10.7769/gesec.v15i4.3739

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

Uma abordagem para descarga com considerações multi-hop em uma

configuração de sobreposição de sinal RSU

Un enfoque para descargar con consideraciones de saltos múltiples en un

ajuste de superposición de señal RSU

Efrem Eladie de Oliveira Lousada1

Fátima de Lima Procópio Duarte Figueiredo2

Abstract

In recent years, significant advancements in vehicle technology have spurred growing interest

in Vehicular Ad hoc Networks (VANETs). This interest is driven by concerns for road safety

and the need to alleviate network congestion, leading to the emergence of Intelligent Transport

Systems (ITS). ITS focuses on improving road traffic management and safety through the

utilization of wireless and mobile network communication technologies. VANETs play a

pivotal role within the realm of ITS, facilitating tasks such as enhancing road safety, traffic

monitoring, and ensuring passenger comfort by mitigating accidents and congestion. These

objectives rely on the timely and accurate delivery of data to vehicle agents and relevant

authorities, facilitated by reliable VANETs and Road Signal Units (RSUs). Achieving this

necessitates identifying optimal routes with minimal distance, high radio access, and quality-

awareness levels. To address these objectives, this study proposes the utilization of the

Congestion Network with Predicted K-means multi-hop RSU algorithm (CN-MHMR) to

enhance vehicular networking and communication. This algorithm facilitates efficient node

transfer from base nodes to destination nodes via the shortest and energy-efficient paths,

thereby enabling viable and reliable vehicular communications. The performance of the

proposed model was evaluated based on various metrics, including energy consumption,

1 Master in Informatics, Instituto Federal de Minas Gerais (IFMG), Ibirité, Minas Gerais, Brasil.

E-mail: efrem.lousada@ifmg.edu.br

2 PhD in Computer Science, Pontifícia Universidade Católica de Minas Gerais (PUC-MG), Belo Horionte, Minas

Gerais, Brasil. E-mail: fatimafig@pucminas.br

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

2

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

throughput, delay, packet delivery ratio, accuracy, precision, and recall values.

Keywords: Vehicular Ad hoc Networks. Road Signal Units. Discrete Packets. Energy

Efficient Routing. Vehicular Communication.

Resumo

Em anos recentes, avanços significativos na tecnologia veicular têm despertado um crescente

interesse em Redes Ad hoc Veiculares (VANETs). Este interesse é impulsionado por

preocupações com a segurança nas estradas e pela necessidade de aliviar a congestão de redes,

levando ao surgimento de Sistemas de Transporte Inteligentes (ITS). ITS foca em melhorar o

gerenciamento do tráfego rodoviário e a segurança através da utilização de tecnologias de

comunicação sem fio e móveis. As VANETs desempenham um papel crucial dentro do

domínio dos ITS, facilitando tarefas como o aprimoramento da segurança rodoviária,

monitoramento do tráfego e garantia do conforto dos passageiros, mitigando acidentes e

congestionamentos. Esses objetivos dependem da entrega oportuna e precisa de dados para

agentes veiculares e autoridades relevantes, facilitada por VANETs confiáveis e Unidades de

Sinalização de Estrada (RSUs). Para alcançar isso, é necessário identificar rotas ótimas com

distância mínima, alto acesso de rádio e níveis de consciência de qualidade. Para abordar esses

objetivos, este estudo propõe a utilização do algoritmo de Rede de Congestionamento com K-

means multi-hop RSU previsto (CN-MHMR) para aprimorar a rede e comunicação veiculares.

Esse algoritmo facilita a transferência eficiente de nós de nós base para nós de destino através

de caminhos mais curtos e eficientes em energia, permitindo assim comunicações veiculares

viáveis e confiáveis. O desempenho do modelo proposto foi avaliado com base em várias

métricas, incluindo consumo de energia, throughput, atraso, taxa de entrega de pacotes,

precisão e valores de recordação.

Palavras-chave: Redes Ad hoc Veiculares. Unidades de Sinalização de Estrada. Pacotes

Discretos. Roteamento Eficiente em Energia. Comunicação Veicular.

Resumen

En años recientes, los avances significativos en la tecnología vehicular han despertado un

creciente interés en las Redes Ad hoc Vehiculares (VANETs). Este interés está impulsado por

preocupaciones sobre la seguridad en las carreteras y la necesidad de aliviar la congestión de

redes, lo que ha llevado al surgimiento de los Sistemas de Transporte Inteligente (ITS). ITS

se enfoca en mejorar el manejo del tráfico rodado y la seguridad a través de la utilización de

tecnologías de comunicación inalámbrica y móvil. Las VANETs juegan un papel crucial

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

3

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

dentro del ámbito de los ITS, facilitando tareas como mejorar la seguridad vial, monitorear el

tráfico y garantizar la comodidad de los pasajeros, mitigando accidentes y

congestionamientos. Estos objetivos dependen de la entrega oportuna y precisa de datos a los

agentes vehiculares y las autoridades relevantes, facilitada por VANETs confiables y

Unidades de Señalización de Carretera (RSUs). Para lograr esto, es necesario identificar rutas

óptimas con distancia mínima, alto acceso de radio y niveles de conciencia de calidad. Para

abordar estos objetivos, este estudio propone la utilización del algoritmo de Red de

Congestión con K-means multi-hop RSU previsto (CN-MHMR) para mejorar la red y la

comunicación vehiculares. Este algoritmo facilita la transferencia eficiente de nodos desde

nodos base hasta nodos de destino a través de caminos más cortos y eficientes en energía,

permitiendo así comunicaciones vehiculares viables y confiables. El rendimiento del modelo

propuesto fue evaluado en base a varias métricas, incluyendo consumo de energía, throughput,

retardo, ratio de entrega de paquetes, precisión y valores de recordación.

Palabras clave: Redes Ad hoc Vehiculares. Unidades de Señalización de Carretera. Paquetes

Discretos. Enrutamiento Eficiente en Energía. Comunicación Vehicular.

Introduction

Vehicular Ad-hoc Networks, known as VANET, are vital in making an Intelligent

Transportation Systems (ITS). Due to increased traffic levels, the topology in the network

changes results as a dynamic challenge for the sparse vehicle distribution and relies in

hindering the scalability of the network. The Multi-Access Edge Computing (MEC) is an

emerging technology that enables computing and storage resources to be moved closer to the

network edge. Access points or base stations are some examples that can be used to reduce

the network latency and to improve data transfer efficiency. MEC can be used in various

applications, including Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I)

communications, which allow vehicles to communicate with each other and with roadside

units (RSUs) [1].

With the increasing demand for services that utilize vehicular networks and the large

amount of data generated by the vehicles, it is necessary to develop strategies to alleviate the

overload in these networks. Offloading is a strategy used to reduce the amount of data/services

transmitted over the network. During this process, data is transmitted using Vehicle-to-Vehicle

(V2V) and Vehicle-to-Road Infrastructure (V2I) communications [2]. The V2V and V2I data

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

4

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

offloading is an important VANET problem to be solved. The use of MEC in this context can

help to reduce the network load and to improve the data offloading performance [3].

Offloading to infrastructure, using V2I communications, can utilize Cellular Base

Stations (CBS) or Roadside Side Units (RSUs). CBSs have higher cost per data transmitted

and should be avoided. RSUs have lower cost for data traffic and are therefore preferred for

V2I offloading [4]. However, while RSUs have lower data traffic costs, they are not widely

available due to deployment costs. In cases where RSUs are not accessible, the offloading

solution must utilize the cellular network [7]. During the offloading from the vehicular

network to the infrastructure, the need for multi-hop routing may arise to ensure that the

information reaches the RSU. The solution must choose the best path between the vehicle

needing to offload data and the RSU. During the selection of the optimal transmission path,

factors such as energy consumption, delivery success probability, and delivery time must be

taken into consideration [8].

In an overlapping multi-RSU environment, multiple RSUs are deployed in an area,

and vehicles can reach multiple RSUs simultaneously. This can be leveraged to improve data

offloading efficiency by allowing vehicles to communicate with the RSU that provides the

best connectivity at any given time [5]. The basic idea of MEC-based data offloading in an

overlapping multi-RSU environment is that vehicles can offload their data to an RSU that can

process the data and forward it to a MEC server located at the network edge. The MEC server

can perform various tasks on the data, such as aggregation, filtering, and data analysis After

that, the MEC can send the processed data back to the RSU for delivery to other vehicles or

to the cloud [6].

To implement MEC-based data offloading in an overlapping multi-RSU environment,

several challenges need to be addressed, such as resource allocation, mobility management,

routing and security [9]. Resource allocation involves allocating computing and storage

resources to vehicles and RSUs based on their requirements and availability. Mobility

management involves ensuring seamless handover of vehicles between RSUs and maintaining

connectivity during the handover process. Security involves ensuring the confidentiality and

integrity of data during transfer and storage. In summary, MEC-based data offloading in an

overlapping multi-RSU environment has the potential to enhance the efficiency and

performance of V2V and V2I communications. However, several challenges need to be

overcome before this technology can be widely deployed in real-world scenarios [10].

In this context, this work presents a solution for offloading considering multiple hops

for transmission and the potential overlap of RSU signals. The proposed study uses a Predicted

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

5

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

K-means Multiple Hop algorithm with the Congestion Network, for finding the efficient MEC

for the vehicular offloading. It can also prevent congestion in the vehicular networks, and also

increase the establishment of the broadcast packets and simultaneously decrease the number

of discrete packets by finding the energy efficient routes for the vehicular communication

[28]. These are done by evaluating the fitness of the RSU which are the intermediate in

providing an uninterrupted vehicular communication. These are in a continuous iteration, until

the energy efficient route is found for the efficient node transfer. These are evaluated using

the appropriate performance metrics, consisting in throughput levels, delay, packet delivery

ratios and levels of energy consumption, for evaluating the energy efficient path identified for

the node transfer from base station to the destination.

The main goal of this paper is effective V2V2I (Vehicle-to-Vehicle-to-Infrastructure)

communication. This is achieved by finding the best offloading ways, considering multiple

RSUs, energy consumption and network congestion. The specific goals were: (1) the CN-

MHMR algorithm implementation to evaluate the RSUs fitness function value ensuring the

vehicular broadcast and the routing for the node transmission using VVR-RSU method and

(2) the overall model performance evaluation through some metrics such as ratio of packet

deliveries, delay time, precision, recall and also the accuracy rates.

The remaining parts of the paper are the following: section III presents some of the

existing models and approaches in vehicle networking and communications, section IV,

explains the solution and the methodology used, section V shows the results. Finally, section

VI provides an overall conclusion and some future work.

Review on Existing Work

Some existing research has focused on improving the effectiveness of vehicular

networking and communications, as listed in this section.

The authors of [17] proposed HetCast, a collaborative data transmission mechanism

for vehicle users that utilizes both cellular LTE networks and IEEE 802.11p. The goal is to

use the IEEE 802.11p network to offload LTE data traffic through RSU-based data

transmission based on data popularity. The proposed method divides data into popular and

unpopular categories and transmits popular data from RSUs, while unpopular data is received

over LTE. The authors also proposed three schemes to establish a global transmission

schedule, including an independent RSU scheduling scheme, a collaborative RSU scheduling

scheme, and a collaborative RSU scheduling scheme with anticipation. Simulation results

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

6

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

showed that the proposed method can effectively reduce the traffic load on the LTE cellular

network and improve download efficiency.

In [18], the authors proposed a QoS-aware offloading scheme to address the problem

of data congestion in cellular networks. The solution offloads delay-tolerant data to vehicular

network components such as vehicles and RSUs. The proposed scheme involves dividing the

delay-tolerant data into equally-sized data blocks (DBs) and storing them in the current

vehicle/RSU or delivering them to the next-hop device. The decision-making process of the

next-hop device is formulated as an optimization problem using a partially observable Markov

decision process (POMDP). Experimental results demonstrate that the authors' scheme

significantly improves performance when compared to other considered schemes. The

effectiveness of the scheme in alleviating congestion in cellular networks is demonstrated by

its ability to offload delay-tolerant data to vehicular network resources. The use of POMDP

for decision-making ensures that the next-hop device makes optimal decisions based on its

partial observation of the system's state.

In [19], the authors proposed an adaptive algorithm for data offloading decision-

making in vehicular networks considering both the cost and delay of cellular base stations

(BS) and roadside Wi-Fi access points (APs). The algorithm was developed taking into

account that users prefer free networks and desire a minimum download waiting time. To

solve the decision-making problem, the proposed algorithm estimated the encounter time

between a vehicle and a Wi-Fi AP, and then the vehicle chose the best download strategy based

on the estimation. However, as the current situation may differ from the original estimate, for

example, if a vehicle encounters a Wi-Fi AP earlier than expected, the proposed algorithm

dynamically adjusted the estimated waiting time and the currently adopted offloading strategy.

Experimental results showed that the proposed method can achieve better user satisfaction

compared to other methods. In summary, the authors' adaptive algorithm considers user

preferences and Wi-Fi AP encounter opportunities, resulting in improved data download

performance in vehicular networks.

The previously mentioned works belong to the traditional V2I scenario, which means

offloading considering only one hop. Our proposed work adopts a VVR (Vehicle-to-Vehicle-

to-Roadside) multi-hop path to perform offloading.

In the traditional V2I scenario, offloading is performed through a direct

communication link between a vehicle and an infrastructure node. This approach is often

limited by the communication range of the infrastructure node and can result in network

congestion and poor performance. In contrast, a multi-hop VVR path defines that vehicles

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

7

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

relay the data to each other until they reach a roadside unit that can send the data to the internet

or other networks. By using a multi-hop approach, the limitations of the traditional V2I

scenario can be overcome, improving network performance. Vehicles can communicate with

each other over longer distances, and the workload is distributed among multiple nodes,

reducing the risk of congestion.

An innovative decision-making framework has been proposed for data offloading to

two types of servers: terrestrial MEC servers and MEC servers mounted on unmanned aerial

vehicles (UAVs), which have distinct characteristics and capabilities [20]. In this proposed

scheme, users have the option to partially offload their data to a complex MEC environment,

considering the latency and the energy requirements. They can use a theoretical prospective

decision-making to maximize perceived utility and reduce time and energy consumption by

terrestrial MEC servers. Numerical results showed that the proposed framework operates well

and the method outperforms other compared methods in the study.

In [21], the authors investigated how to improve MEC access capability and increase

spectrum utilization efficiency by studying the task offloading and resource allocation

problem. The main idea proposed was to maximize the MEC processing capacity as an

optimization objective in a multi-user, multitask, and multi server scenario. The proposed

approach divided the mixed-integer nonlinear programming (MINLP) problem into a resource

allocation problem and a task allocation problem. Furthermore, the resource allocation

problem was divided into computation resource allocation and communication resource

allocation. To deal with these problems, the authors proposed a low-complexity suboptimal

matching algorithm for subchannel allocation to maximize task offloading efficiency.

The work developed in [22] addressed the joint allocation of spectrum, computation,

and storage resources in a multi-access edge computing (MEC) based vehicular network. Two

typical MEC architectures were considered, and multidimensional resource optimization

problems were formulated. The authors of the work utilized the optimization problems with

the aim of maximizing the number of tasks offloaded while meeting quality of service (QoS)

requirements, considering the limited quantities of available spectrum, computation, and

storage resources. To achieve the ideal spectrum allocation between base stations and the

optimal allocation of spectrum, computation, and storage among vehicles, the authors

transformed the optimization problems into a Deep Reinforcement Learning problem and

proposed an algorithm based on deep deterministic policy gradients to solve them.

While the previously presented work focused on offloading through MEC servers, our

proposed work focuses on data offloading through RSUs, transferring data traffic from the

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

8

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

4G/5G cellular network to the IEEE 802.11p vehicular network with the assistance of MEC

servers. Unlike the distributed computing approach used in traditional Vehicular Ad Hoc

Network (VANET) to find the k-hop V2V path, our proposed work adopts a centralized

computing approach, which can be achieved with the help of the MEC server.

In [23], a proposed method for vehicular data offloading, called VVR offloading path

selection, based on the destination's retransmission time limit, utilizes a Mobile Edge

Computing (MEC) approach instead of the traditional V2I offloading method. The proposed

method aims to perform data offloading using RSU X, n hops away, by selecting the most

suitable VVR data offloading path. To achieve this, the MEC server receives periodic context

reports from vehicles and utilizes detection and reduction (DAS), detection and extension

(DAE), and path recovery mechanisms to construct and maintain the VVR data offloading

path. The DAS and DAE mechanisms use OA reselection to reconstruct the VVR data

offloading path when the originating vehicle enters or exits the signal range of the RSU ahead

or behind it. Additionally, the path recovery mechanism is triggered to maintain VVR data

offloading if a vehicle in the relay route deviates from the VVR data offloading path. In

summary, the proposed method aims to enhance vehicular data offloading by using an MEC-

based approach that selects the most suitable VVR data offloading path and maintains

offloading using RSU X whenever possible.

The authors of the article [24] proposed a novel method called Delay-Constrained k-

hop Limited Utility-based VVR Data Offloading Path Construction Method (DC-KUPC).

This method is based on MEC and is used to find the best VVR data offloading path between

an originating vehicle and an RSU, considering a time period with a delay constraint. To

implement this method, the MEC server receives a context report from the originating vehicle

at time point tr and then calculates all possible VVR data offloading paths between the

originating vehicle and RSU Y within a time period of tr + T. T represents the delay-

constrained time period, and during this time, the MEC server searches for all candidate VVR

data offloading paths that satisfy the delay constraint.

A utility function is then used to derive the quality of each candidate VVR data

offloading path. The utility function takes into account various factors such as path length,

bandwidth, and signal intensity. Finally, the DC-KUPC method selects the VVR data

offloading path with the highest utility value as the best VVR data offloading path. Compared

to the lifespan-based method proposed in [25], which considers only the lifespan of the path

without considering the network quality of the path, the proposed DC-KUPC method achieves

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

9

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

better data offloading performance. Simulation results demonstrate that the proposed utility-

based method can significantly improve the data offloading performance in VVR systems.

Considering offloading in a multi-RSUs and n-hop environment, three works are

presented in the literature. The following three described works do not consider energy

consumption for calculating the best route, nor the time it takes for the data to be transmitted.

They also do not consider the transmission time of the VVR data offloading path, thus making

them unsuitable for real-time solutions. One of these papers presents a method called

Predicted K-hop-limited Multi-RSU-considered (PKMR) for vehicle-to-vehicle-to-roadside

unit (VVR) data offloading in a multi-access edge computing (MEC) environment. The

method utilizes a Software Defined Network (SDN) controller within the MEC server to

manage the offloading process. By considering the predicted paths and network conditions of

vehicles and roadside units (RSUs), the PKMR method selects the most suitable VVR data

offloading path. The performance evaluation shows that PKMR outperforms traditional self-

offloading methods. The proposed method addresses challenges related to multi-RSU

deployment, RSU signal overlap, and data offloading [26]. But it does not consider energy

consumption for calculating the best route, nor the time it takes for the data to be transmitted,

thus making it unsuitable for real-time solutions.

In the work presented in [27], the authors propose a multi-user and multi-RSU system

architecture based on the Internet of Vehicles (IoV) using SDN. In order to reduce the

offloading delay in IoV, a joint approach is proposed to optimize the offloading rate,

offloading decisions, and resource allocation. The solution proposes the use of RSUs for

computing vehicular network activities in order to balance the utilization of computational

resources between the vehicular network and RSUs. Although the work proposes the use of

multi-RSUs, it does not address the case when the signals from the RSUs overlap. It also does

not handle the reselection of a new path if there is a disconnection of vehicles that are part of

the offloading path.

Some of the core concerns in the existing methods are the following:

● some papers work with the traditional V2I scenario, which means offloading

considering only one hop;

● some of the metaheuristic methods can be implemented in making the resource

utilization to maximum levels and can also be distributed to other nodes [28];

● some of the previously presented work focused on offloading just through MEC

servers;

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

10

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

● alternative ways in exploring and defining the critical areas are reduced as rectangular

regions. They are not applicable in some of the scenarios. Also, an extended approach

can be made in finding other applications for enhancing the content viability and in

reducing the backhaul traffic [7];

● the presented works do not take into account energy consumption as well as the time

it takes for the data to be transmitted, thus making it impractical for real-time solutions.

Proposed Methodology

This section presents the complete work methodology. The wireless vehicular

communication uses RSU´s. These RSU´s are efficient in making short range communications

which are operated in a short spectrum range. The proposed method uses the VVR-RSU

signalling protocol. In the initial levels, the system model consists of VVR, which is the path

chosen. It is used in the offloading of the destination vehicle. This is done using the proposed

CN-MHMR, with the Vehicle to Vehicle - RSU. This combination is known to be RSU-VVR.

The data is transferred from one node to another that receives the packet via the efficient

energy routes. These data packets are transferred from the node (vehicle) using the nearest

RSU. This is done for finding the efficient route for the effective data transfer. Using the

energy efficient routes results in energy efficient data transfer. Some of the parameters

analysed comprises the bandwidth, the packet delivery ratio, the energy consumption and the

throughput.

The diagrammatic representation of the proposed method is given in Figure 1. It can

be seen in the diagram that the first step in the CN-MHMR module, the first thing to do is the

agent's initialization. When a node receives a data packet, the best route is calculated based

on the external fitness evaluation of the RSU signal, energy estimation and the best link quality

between vehicles, calculated using complex network metrics. The retransmission probability

is calculated for each data packet received, forming a loop until the broadcast decision is

carried on. When the packet is broadcasted, it is discarded by the CN-MHMR. The

performance analysis is made over all the collected data: energy consumption, bandwidth,

throughput, delay and packet delivery ratio are some of the possible metrics to be presented.

All the modules and methods will be detailed explained in the next sections.

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

11

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

Figure 1

Overall flow of the proposed methodology

3.1 Proposed Method CN-MHMR

The proposed approach comprises four stages as outlined below; (a) Initialization

Phase; In this stage the vehicle Vs utilizes the network while the MEC server endeavors to

find a V2V2I VANET offloading route, for Vs. (b) Shrinking Phase; During this phase vehicle

Vs moves towards an RSU that provides signal coverage gradually reducing the length of the

V2V2I VANET offloading path. This reduction is achieved by decreasing the number of hops

in the V2V2I path. (c) Self offloading Phase; At this point vehicle Vs is within range of an

RSUs signal coverage. Can independently perform offloading without relying on vehicles. (d)

Extending Phase; Throughout this phase vehicle Vs moves from the RSUs signal coverage

area resulting in an increase in the number of hops required for V2V2I communication. In

phases 2 and 4 there may be instances where it becomes necessary to reconstruct the path. If

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

12

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

such a situation arises the algorithm returns to phase 1 to recalculate routes and establish links,

between network nodes.

3.1.1 The initialization phase

Initially, the vehicle Vs uses the 4G/5G cellular network to communicate with its peer

Vp on the infrastructure Internet side. When the MEC server receives the periodically reported

context from vehicle Vs at time point Tc, it is triggered to determine if there are one or more

V2V2I offloading paths for the vehicle Vs. If the MEC server finds a V2V2I offloading path

for Vs, it calculates all potential V2V2I paths that may exist during the interval [Tc, Tc + t]

and selects the best one as the V2V2I offloading path, generated at time point T0, where To is

within the interval [Tc, Tc + t], for Vs. Subsequently, the MEC server sends messages to the

constituent vehicles of the corresponding V2V2I offloading path to enable the V2V2I VANET

offloading session for Vs at time point T0.

The delay-constrained time length t is set to be smaller than or equal to the

corresponding vehicle's periodically reported context time period because the MEC server can

only receive the next reported context after one periodic reporting time period. During the

offloading session, all constituent vehicles of the V2V2I offloading path synchronously report

their contexts to the MEC server, which can be achieved by aligning the time with GPS time.

After receiving the reported contexts from the constituent vehicles of the V2V2I offloading

path, the MEC server calculates the time for each V2V link and the V2I link, which represents

the offloading agent's duration inside RSU's signal coverage, to update the remaining

offloading time of the corresponding V2V2I offloading path.

The initial step involves identifying the start time points and end time points for all

links. It is based on the location speed of every node pair (p,q) in the environment (NodeSet),

the geo-distance, signal range of the on-board unit and fitness function for energy. To achieve

this, Procedure 1 in Figure 2 is executed, to derive the start time point and end time point of

each link, which was proposed in [26]. Consider the position and speed of each pair of vehicles

p and q, where both p and q are part of the NodeSet at time point t0. These can be represented

as the coordinates [Posx(p),Posy(p)] and [Posx(q),Posy(q)], and the velocities

[Velx(p),Vely(p)] and [Velx(q),Vely(q)], which can be communicated in a range R of 300m

(equal to IEEE 802.11p OBU).

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

13

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

Figure 2

Procedure Find All of the Links’ Connected Time Intervals

Procedure 1: Find All of the Links’ Connected Time Intervals

1: for each Link (p, q) do

2: ([𝑝𝑥(𝑝) − 𝑃𝑥(𝑞) + 𝑡(𝑉𝑥(𝑝) − 𝑉𝑥(𝑞))]

2

+

[𝑃𝑦(𝑝) − 𝑃𝑦(𝑞) + 𝑡(𝑉𝑦(𝑝) − 𝑉𝑦(𝑞)]2) − 𝑅 ∗ 𝑅 = 0

3: End for

If there are no solutions, to the equation used in Procedure 1 it means that the shortest

distance between object p and object q is greater than R. Consequently Link (p, q) is

considered invalid and removed from the analysis. If there is one solution it indicates that Link

(p, q) breaks immediately and needs to be removed. However if there are two solutions it

implies that Link (p, q) exists for a time interval where one solution represents the starting

time point and the other solution represents the ending time point of Link (p,q). In cases where

an infinite number of solutions exist it suggests that Link (p,q) can persist indefinitely. This

means that objects p and q have the driving speed and direction while their initial distance is

not greater than R.

Once we have identified all start time points and end time points for each link we can

determine the suitable VVR data offloading path. For every link, on each path we utilize a

success probability function (SP).

𝑆𝑃 = 𝑄𝑙 ∗ 𝐹𝑓 ∗ 𝐶𝑝

(1)

Each of the components of the formula will be specified below. The quality function

takes into account several factors to determine the quality of a V2V2I offloading path. These

considerations are as follows: (i) A longer lifetime of the V2V2I path leads to a higher utility

value, indicating its desirability, (ii) If there are more vehicles within the signal coverage of a

constituent vehicle, it may result in potential transmission collisions, leading to a lower utility

value due to higher backoff values required for channel access, (iii) The total number of

vehicles within the signal coverage of each constituent vehicle influences the path's utility

value, indicating that a more congested road results in a lower utility value. A quality function

that can derive the VVR data offloading path’s quality is defined as follows, for which the

used variables are defined and explained in Table 1. Ql is based in [24].

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

14

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

𝑄𝑙 =

(

2)

Table 1

Variables Used in the Quality Function

Parameter

Description

Toffloading

The lifetime of the V2V2I offloading path, which is the minimum link's

connected time of the constituent links in the V2V2I offloading path.

Ni

The number of vehicles that are in the signal coverage of the offloading path's

constituent vehicle V.

Nmax

Max (N1, N2, ..., Nn).

Ntotal

Sum of (N1, N2, ..., Nn).

n

The hop count in the V2V2I offloading path, including the V2I link from the

offloading agent to RSU.

The fitness function (Ff), Figure 3, has been designed to evaluate the fitness value of

a vehicle's position in a 2D space relative to its counterpart. It takes two parameters as input:

energy, which represents the initial energy level of the vehicles and distance, which is a vector

containing the distances between the vehicle under consideration and other vehicles in the

environment. The function proceeds with calculating three weighted components of the fitness

value. Firstly, it calculates the average distance (x1) between the vehicle under consideration

and all other vehicles within its sensing range. Secondly, it evaluates x2, which involves

comparing the initial energy (energy) to the product of energy and the number of vehicles (Cn)

within the sensing range. As the product simplifies to the original energy value, x2 essentially

becomes 1. Finally, x3 is obtained as the inverse of the number of vehicles within the sensing

range. Using predefined constants alpha1, alpha2, and alpha3, the function combines the three

components (weighted accordingly) to generate the final fitness value. These coefficients

(alpha1, alpha2, and alpha3) allow us to adjust the importance of each factor in the fitness

calculation.

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

15

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

Figure 3

Algorithm fitnessFunc

Algorithm fitnessFunc (energy, distance)

1: alpha1 ← 0.2

2: alpha2 ← 0.5

3: alpha3 ← 1 - alpha1 - alpha2

4: Cn ← sum (distance < Sensing_range)

5: x1 ← sum (distance) / Cn

6: x2 ← ((energy * Cn) / Cn) / energy

7: x3 ← 1 / Cn

8: fitness_val ← (alpha1 * x1) + (alpha2 * x2) + (alpha3 * x3)

9: end

The efficiency of a link in performing transmission is determined by a probability

function that considers factors such as vertex degree, topological overlap, and edge

persistence. This function calculates the transmission success probability (Cp), which is then

compared to a threshold value derived from an exponentially weighted moving average. If Cp

exceeds the threshold, the link is penalized and may be ranked lower. The selection of

components for calculating the retransmission probability is based on the significance of a

specific vehicle in the network's packet retransmission process. The following complex

network metrics are used [29]:

● vertex degree: Vertices with a higher degree are more active in the network and can

serve as channels for information exchange;

● topological overlap: Depicts the tendency for vertices to have shared neighbors.

Therefore, the fewer shared neighbors, the higher the possibility of reaching a larger

number of vehicles;

● edge persistence: Defines the number of times two vertices encounter each other within

the same time window. Vehicles that encounter each other in the same time window

have a higher probability of successful information exchange.

The value of p is calculated by the product of the probability associated with the Vertex

Degree (pGV), the probability associated with the Topological Overlap (pST) and the

probability associated with Edge Persistence (pPA), formulated by the Equation 3:

𝐶𝑝 = 𝑝𝐺𝑉 ∗ 𝑝𝑆𝑇 ∗ 𝑝𝑃𝐴

(3)

To compute the pGV for a particular vehicle, it requires information about the vertex

degree of all its neighboring vehicles, which is kept and updated in its neighbor table. The

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

16

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

vehicle then calculates the ratio between its own vertex degree (gV) and the sum of the vertex

degrees of its neighbors (gVV).

𝑝𝐺𝑉 =

𝑔𝑉

𝑔𝑉𝑉

(4)

While disseminating data, it is preferable to choose vehicles with the fewest common

neighbors. The calculation of pST is determined by subtracting one from the quotient of the

number of common neighbors (nVC) divided by the sum of the transmitter's neighbors (nVT)

and the receiver's neighbors (nVR).

𝑝𝑆𝑇 = 1 −

𝑛𝑉𝐶

𝑛𝑉𝑇+𝑛𝑉𝑅

(5)

The transmitting vehicle takes into account the two seconds leading up to the receiver

selection in order to compute the pPA value. If the link was consistently active during those

two seconds, pPA is assigned a value of 1. If it was active only in the last second, it receives

a value of 0.5. If there was no link in the last second, it is assigned a value of 0, preventing

the vehicle from being selected as a receiver. After calculating Cp, it is essential to verify if

the value surpasses the threshold. If it exceeds the threshold, it undergoes a 50% penalty, and

the penalized link is moved to the last position.

The threshold calculation is based on the methodology employed by the E-probT

protocol outlined in [30]. The initial threshold value, denoted as avgThr, is derived by

computing the average Cp for the three vehicles farthest from the transmitter. To refine this

value, it undergoes a weighted exponential moving process that takes into account both past

and current values for computing new thresholds. This process aims to mitigate potential

discrepancies and errors in the ongoing estimation. Similar to TCP's timeout mechanism, it

involves a smoothing of the values curve. Using avgThr, the estimated threshold value,

referred to as estThr, is determined as shown in Equation 6, which represents the calculation

for the estimated threshold.

𝑒𝑠𝑡𝑇ℎ𝑟 = (1 − 𝛽) ∗ 𝑒𝑠𝑡𝑇ℎ𝑟 + 𝛽 ∗ 𝑎𝑣𝑔𝑇ℎ

(6)

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

17

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

The value of the used β constant is the same as the work [39][30], recommended value

for TCP. Thus, β = 1/8 = 0.125 [38]. Besides the estimating, it is necessary to calculate the

threshold variation, called varThr. This calculation is performed based on the Equation 7.

𝑣𝑎𝑟𝑇ℎ𝑟 = (1 − 𝛾) ∗ 𝑣𝑎𝑟𝑇ℎ𝑟 + 𝛾 ∗ |𝑎𝑣𝑔𝑇ℎ𝑟 − 𝑒𝑠𝑡𝑇ℎ𝑟|

(7)

The value of the used γ constant is the same as the work [39][30], recommended value

for TCP. Thus, γ = 1/4 = 0.250 [38]. Finally, the final threshold value, called endThr, is

calculated by the sum of its estimated value (estThr) and the variance (varThr) value. Equation

8 shows the calculation for the final threshold value.

𝑒𝑛𝑑𝑇ℎ𝑟 = 𝑒𝑠𝑡𝑇ℎ𝑟 + 𝑣𝑎𝑟𝑇ℎ

(8)

With the defined probability of success function, the candidate paths for offloading

can be determined. The InitContenderSet, Figure 4, algorithm aims to initialize the set of

candidate paths for finding the best path from Vroot to other nodes within a node set (NodeSet)

and a time point set (TimeSet). Firstly, the set of candidate paths (ContenderSet) is initialized

as empty. Then, the algorithm iterates through each node v in the NodeSet and each time point

t in the TimeSet. During the iteration, the algorithm checks if there is an existing link between

Vroot and node v that is active at time point t. If such a link exists, a new candidate path is

created, starting at Vroot and ending at v, at time point t. The new candidate path is then

assigned with relevant information, including the start node, end node, postponed time,

lifetime, quality of the path calculated using a specific function, and hop count, which is set

as 1 since it is a one-hop path from Vroot to v. This candidate path is added to the set of

candidate paths (ContenderSet) for further consideration. After completing the iterations

through the nodes and time points, the algorithm returns the ContenderSet containing all

possible paths from Vroot to nodes in the NodeSet that are active at their respective time

points. This set will be used in subsequent processes to determine the best path from Vroot to

the final destination node or Roadside Unit (RSU).

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

18

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

Figure 4

Algorithm InitContenderSet

Algorithm InitContenderSet(Vroot, NodeSet, TimeSet)

1: ContenderSet ← ∅

2: linksToVroot ← Links from Vroot to NodeSet

3: for each v in NodeSet do

4: for each t in TimeSet do

5: for each link (p, q) in linksToVroot do

6: if tstart(link) ≤ t and tend(link) > t then

7: NewContenderPath ← Create new path from Vroot to v

starting at t

8: End(NewContenderPath) ← v

9: PostponedTime(NewContenderPath) ← t

10: Lifetime(NewContenderPath) ← tend(link) - t

11: Quality(NewContenderPath) ←

ComputeQuality(NewContenderPath)

12: HopCount(NewContenderPath) ← 1

13: add NewContenderPath to ContenderSet

14: end if

15: end for

16: end for

17: end for

18: return ContenderSet

Algorithm CBOP (Constrained Best-Offloading Path), Figure 5, is designed to find the

best data offloading path in a V2V2I (Vehicle-to-Vehicle-to-Infrastructure) vehicular network.

The goal is to identify the path that offers the best communication quality while considering

the delay constraint imposed by the system. The algorithm begins by initializing a set of

candidate paths (ContenderSet) using the InitContenderSet function, which creates all

possible one-hop paths from the root vehicle Vroot. It then enters a loop (line 2) to iterate over

the ContenderSet. Within the loop, the algorithm calculates the quality of all offloading paths

in ContenderSet (line 3) and selects the path that has the maximum quality value as Pathmax

(lines 4 to 10). If the Pathmax path reaches the RSU (Infrastructure Service Unit), the loop is

interrupted as the best path has been found, and there is no need to continue the search. If

Pathmax does not reach the RSU, the ContenderSet is updated to contain only promising paths

to continue the search (line 15), and the Pathmax path is removed from the set to avoid

unnecessary repeated calculations (line 16). The algorithm continues to iterate until there are

no more candidate paths in the ContenderSet. It then returns the Pathmax path, which is

considered the best data offloading path within the constraints of delay and quality.

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

19

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

Figure 5

Algorithm Constrained Best-Offloading Path

Algorithm CBOP (Vroot, Tpred−limut, Linklimit, k)

1: ContenderSet ← InitContenderSet(Vroot, NodeSet, TimeSet)

2: Pathmax ← null

3: maxQuality ← -∞

4: while ContenderSet is not empty do

5: for each Path in ContenderSet do

6: quality ← calculateQuality(Path)

7: if quality > maxQuality then

8: maxQuality ← quality

9: Pathmax ← Path

10: end if

11: end for

12: if End(Pathmax) = RSU then

13: break

14: end if

15: ContenderSet ← RefreshContenderSet(Pathmax, ContenderSet, k)

16: ContenderSet.remove(Pathmax)

17: end while

18: return Pathmax

3.1.2 The shrinking phase

The main goal is to maintain the V2V2I offloading path for as long as possible. When

the MEC server receives the periodically reported context from the offloading agent at the

current time point Tc, it knows that the offloading agent will be out of the RSU's signal

coverage when its next periodically reported context is received, indicating that it is leaving

the RSU's signal coverage. To address this, the shrinking algorithm is triggered to find a new

offloading agent and update the V2V2I offloading path. The MEC server selects the updated

path from all the potential paths that may exist during the interval [Tc, Tc + t]. The sub-path

of the V2V2I offloading path that lies outside the RSU's signal coverage is referred to as sub-

pathout-RSU.

To ensure that the delay-constrained time length t does not exceed the lifetime of sub-

pathout-RSU, the delay constraint is defined to be smaller than or equal to the lifetime of sub-

pathout-RSU. If a new offloading agent and/or an updated V2V2I offloading path cannot be

found, the MEC server will inform vehicle Vs to switch back to the cellular network, and the

offloading session will be terminated. However, if vehicle Vs remains within the RSU's signal

coverage, indicating that the offloading path is still intact, the shrinking phase is concluded,

and vehicle Vs can continue with the offloading on its own.

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

20

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

3.1.3 The self-offloading phase

When vehicle Vs is within the signal coverage of the RSU, it can directly communicate

with its Vp through the RSU. During this phase, the MEC server continuously calculates the

time duration that vehicle Vs stays within the RSU's signal coverage, based on the periodically

reported context from Vs. The MEC server receives the context from Vs at a certain time point

Tc, and it knows that vehicle Vs will leave the RSU's coverage when its next context is

received, indicating that vehicle Vs is moving out of the RSU's signal coverage. This transition

triggers the start of the extending phase.

3.1.4 The extending phase

When the source vehicle Vs or the offloading agent of the extending phase moves out

of the RSU's signal coverage, the extending algorithm is activated. At the current time point

Tc, the MEC server receives the periodically reported context of the offloading agent, and it

predicts that the source vehicle or the offloading agent will leave the RSU's signal coverage

upon receiving its next context. This event triggers the extending algorithm, which aims to

find a new offloading agent and an updated V2V2I offloading path. The MEC server evaluates

all potential candidates that may be available during the interval [Tc, Tc + t]. The current

V2V2I offloading path from vehicle Vs to the offloading agent, which is inside the RSU's

signal coverage but is leaving it, is referred to as sub-pathc-offloading. To ensure that the

delay-constrained time length t is reasonable and doesn't exceed the lifetime of sub-pathc-

offloading, t is defined to be smaller than or equal to the lifetime of sub-pathc-offloading. If

the MEC server cannot find a new offloading agent and/or an updated V2V2I offloading path,

it will notify vehicle Vs to switch back to the cellular network, and the offloading session will

be terminated.

3.1.5 Reconstruct the offloading path

When a V2V2I offloading path experiences an unexpected breakage, the path recovery

algorithm is triggered to restore the connection. Let's consider a sub-path Vb – Vy –Va within

the V2V2I offloading path for vehicle Vs. If Vb is unable to transmit data to Va, for example,

because Vy has left the road, resulting in the breakage of the V2V2I offloading path, we denote

this event as the broken time point Tc. The path recovery algorithm attempts to find a new

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

21

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

vehicle Vx that can repair the V2V2I offloading path, or it looks for an opportunity for Vb to

reconnect with Va within the interval of [Tc, Tc + t]. Here, sub-patha represents the current

offloading sub-path from Vs to Va, and sub-pathb-OA represents the current offloading sub-

path from Vb to the offloading agent. To ensure that the delay-constrained time length t is

reasonable and doesn't exceed the lifetimes of sub-patha or sub-pathb-OA, t is defined to be

smaller than or equal to the minimum lifetime between sub-patha and sub-pathb-OA. If the

path recovery algorithm fails to repair the V2V2I offloading path, the MEC server will inform

vehicle Vs to switch back to the cellular network, and the V2V2I offloading session will be

terminated.

Results and Discussion

The proposed method using the VVR - with the CN-MHMR algorithm is used in

finding the energy efficient routes for effective data transfer from one node to the other. The

complete results which were obtained after the placement of the projected process is presented

in the respective section.

4.1 Simulation Results

This subsection contains the system model which is deployed in the particular study,

the graph depicting the routing path and the method of selecting the cluster head for the

effective data transfer.

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

22

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

Figure 6

System model for proposed methodology

Figure 7

Cluster head selection by CN-MHMR

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

23

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

Figure 8

Routing path by CN-MHMR

Figures 6 and 7 depict the system model, and the cluster head selection which is energy

efficient making less discrete packets and increased rates of broadcast packets. The system

model represented is deployed for the proposed study used in finding the effective route for

node transfer. Whereas cluster heads selection is found using the proposed CN-MHMR model

for effective VVR (Vehicle-to-Vehicle-to-Roadside) multi-hop path to perform offloading.

Figure 8 represents the routing path found using the proposed algorithm for the effective data

transfer from one node to the other. This is used in making easier and effective means of

communication facilitating the effective node transfer. Only after the effective selection of the

cluster head the energy effective route is selected and is optimised for effective node transfer

from the base station to the destination node. The groups inside each of the clusters represent

the individual groups, which are considered as one single group. These clusters are instructed

in following one effective route of the node transfer.

The departure time intervals for vehicles vary depending on the density of vehicles. In

high-density situations, the departure time period is 0.8 seconds, in medium-density situations

it is 1.25 seconds, and in low-density situations, it is 2 seconds. Each vehicle starts from a

fixed position and exhibits random driving behaviors. Vehicles have the option to drive away

using the available road intersections. When a vehicle is on the road, it sends a context report

to the MEC server at regular intervals. In normal situations, the report is transmitted every 2

seconds. However, if the vehicle is part of a VVR (Vehicle-to-Vehicle-to-Roadside) data

offloading path, the report is transmitted every 1 second. The data flow between an

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

24

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

infrastructure entity on the roadside and the source vehicle can be established through either

an RSU (Roadside Unit) using an IEEE 802.11p network or a BS (Base Station) using an LTE

cellular network. The data is then transmitted to the source vehicle. Table 2 provides an

overview of the simulation environment parameters and their respective values.

Table 2

Parameters used in the simulation environment

Parameter

Value

Simulation time length

300s

Vehicle´s departure time period

(high/medium/low)

0.8s/1.25s /2s

Vehicle´s velocity limit

60km/h

Context reporting period

1s/2s

RSU signal range

300m

4.2 Performance Analysis

This subsection is used in representing the overall performance exhibited by the

proposed method, for effective node transfer from one node to the other and from the base

node to the destination node. The complete section relies on depicting the fraction of data

offloading levels at different hop levels for the routing in wireless communications.

Four construction strategies for the vehicular network's data offloading path are

evaluated in this study. The first one is the proposed method, a k-hop-limited multi-RSU VVR

data offloading approach considering quality of service, energy consumption, and time-

extended prediction mechanism. The second method has the same characteristics but does not

consider energy consumption for route calculation or delivery probability. The third method

is a k-hop-limited multi-RSU VVR data offloading approach considering quality of service,

without using the time-extended prediction mechanism. The fourth comparison method does

not allow for RSU handoff if the link with the RSU is lost.

About the four strategies, four performance metrics were evaluated: (I) Data loss rate:

This metric quantifies the proportion of lost data in the VVR data offloading path, divided by

the total amount of data transmitted through RSUs. (II) Data offloading fraction: This metric

represents the proportion of offloaded data that were directed through RSUs, divided by the

total amount of data directed through both the BS and RSUs. (III) Successful data offloading

fraction: This metric indicates the proportion of data that were successfully received by the

source vehicle through the VVR data offloading path, divided by the total amount of data

transmitted through both the BS and RSUs. (IV) Number of data offloading sessions: This

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

25

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

metric represents the average number of data offloading sessions that existed during the

simulation.

Figure 9

Data loss rate at 10 hop

Figure 10

Data loss rate at 8 hop

The graphics, in Figures 9, 10, 11 and 12 illustrate how the rate of data loss varies with

numbers of hops. Several factors contribute to the data loss rate. Firstly when there are

vehicles on the road the data loss rate tends to increase. This happens because higher vehicle

density leads to data transmission between vehicles, which can result in collisions if the data

is sent through the VVR data offloading path. This collision effect contributes to a data loss

rate. Secondly as the number of hops increases so does the data loss rate. This is because a

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

26

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

longer VVR data offloading path means that the transmitted data needs to pass through VV

links making it more vulnerable, to losses.

Figure 11

Data loss rate at 4 hop

Figure 12

Data loss rate at 2 hop

Thirdly, the data loss rate in methods 1 and 2 is slightly lower than in method 3. This

means that methods adopting the time-extended prediction mechanism in conjunction with

energy optimization have a lower data loss rate. This is because the time-extended prediction

mechanism helps the MEC server find more candidate paths. Additionally, a candidate path

that consumes less energy requires less effort for package delivery, increasing the probability

of successful delivery. Therefore, the selected VVR data offloading path can provide better

quality of service (QoS), resulting in a lower data loss rate. Lastly, the data loss rate in methods

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

27

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

1 and 2 is slightly lower than in method 3. This means that methods allowing the OA to

perform RSU handoff in regions with overlapping RSU signals, i.e., in environments with

multiple RSUs, have a lower data loss rate.

Figure 13

Data offloading fraction at 2 hop

Figure 14

Data offloading fraction at 4 hop

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

28

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

Figure 15

Data offloading fraction at 8 hop

Figure 16

Data offloading fraction 10 hop

The Figures 13, 14, 15 and 16 graphics deliberate the overall data fraction of the

proposed model in the wireless sensor, for a data transfer from one node to the other. The data

offloading fraction is the amount of offloading data that was directed through RSUs divided

by the amount of data that was directed through both the BS and RSUs. There are four factors

that affect the data offloading fraction. Firstly, a higher vehicle density leads to an increased

data offloading fraction. This is because a VVR data offloading path is more likely to be

established in situations with a higher vehicle density. Secondly, the data offloading fraction

is higher when the system has a larger hop count limit. This is because a larger hop count limit

allows the source vehicle to initiate data offloading earlier, enabling the establishment of a

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

29

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

data offloading session at an earlier stage. As a result, the total time for offloading data traffic

is longer, leading to a higher data offloading fraction.

Thirdly, the data offloading fraction from method 1 and 2, which employs the time-

extended prediction mechanism, is always higher than other methods. This relationship is due

to the fact that the time-extended prediction mechanism offers two advantages. The first

advantage is the potential utilization of VVR data offloading paths with shorter lifetimes.

Another advantage of the time-extended prediction mechanism is the ability to initiate a data

offloading path at an earlier time point. This advantage arises from the detection of potential

VVR data offloading paths that may not exist at the time when the MEC server receives the

source vehicle's periodic context report but will become available in the future, specifically

between two consecutive time points of context reports.

The fourth factor that affects the data offloading fraction is the deployment of multiple

RSUs and RSU handoff. In methods 1 and 2, where both the OA (Offloading Agent) and the

source vehicle can utilize RSU handoff in regions where multiple RSUs have overlapping

signals, the data offloading fraction in a multi-RSU environment is higher compared to other

methods, where the OA does not perform RSU handoff. This relationship occurs because if

the MEC server cannot find an alternative VVR data offloading path using the OA reselection

scheme, the method that incorporates RSU handoff can maintain the VVR data offloading

session by transitioning to subsequent RSUs along the original path. As a result, enabling both

the OA and the source vehicle to perform RSU handoff extends the duration of the VVR data

offloading session, resulting in a higher data offloading fraction.

Figure 17

Successful Data offloading levels at 2 hop levels

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

30

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

Figure 18

Successful Data offloading levels at 4 hop levels

Figure 19

Data offloading levels at 8 hop levels

Figure 20

Data offloading levels at 10 hop levels

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

31

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

The same successful data of lading levels using different density of the node hop are

depicted in the Figures 17, 18, 19 and 20 graphics. The successful data offloading fraction is

influenced by both the data offloading fraction and the data loss rate. Observing the evaluation

results, can be seen that the successful data offloading fraction is proportional to the vehicle

density. This relationship occurs because higher vehicle density leads to a higher data

offloading fraction, outweighing the impact of the increased data loss rate. Additionally,

increasing the hop count limit in each vehicle density situation increases the successful data

offloading fraction. This is due to the fact that the effects of enabling the OA to perform RSU

handoff in a multiple-RSU environment and utilizing the time-extended prediction mechanism

have a more significant impact than the data loss rate. As a result, the successful data

offloading fraction is greater in high vehicle density situations compared to low vehicle

density situations.

Figure 21

Number of Session at 2 hop

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

32

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

Figure 22

Number of Session at 4 hop

Figure 23

Number of Session at 8 hop

Figure 24

Number of Session at 10 hop

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

33

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

The evaluation results depicted in Figures 21, 22, 23 and 24 illustrate that the number

of data offloading sessions varies based on both the hop count limit and vehicle density. When

considering a hop count limit (2 hop 4 hop limited) it is observed that the medium vehicle

density scenario exhibits the highest number of sessions compared to situations, with low and

high vehicle densities. This can be attributed to the opportunities for the MEC server to find

VVR data offloading paths in scenarios with medium vehicle density resulting in paths with

shorter lifetimes and consequently leading to a higher number of sessions. On the hand when

dealing with a hop count limit (8 hop and 10 hop limited) the number of sessions shows an

inverse relationship with vehicle density. As vehicle density increases there is a decrease in

the number of data offloading sessions. This can be explained by considering that in scenarios

with a hop count limit there are possibilities, for selecting longer lifetime VVR data offloading

paths instead of shorter ones. Consequently this prolongs the lifespan of a session while

reducing the overall number of sessions.

To summarize when we raise the hop count limit from 2, to 4 it leads to a number of

offloading sessions in scenarios with a hop count limit. On the hand increasing the hop count

limit from 8 to 10 results in a decrease in sessions for situations, with a hop count limit.

Conclusion

In a multi-RSU and multi-vehicle environment, it is necessary to develop solutions

that consider all possibilities to reduce packet loss and network data traffic. CN-MHMR has

demonstrated that utilizing parameters related to link quality between vehicles and the energy

factor as inputs to select the best data traffic path should be embraced by existing and future

solutions. Until this present work, these metrics were not taken into consideration. Compared

to solutions presented in the literature up to this point, CN-MHMR decreased the packet loss

rate, increased the success in offloading, and maintained a stable number of sessions, just like

previous solutions. When analyzed in comparison with well-established generic solutions

from the literature, CN-MHMR showed improvements in all the metrics analyzed. Future

work may include testing with new mobility models, as well as the adoption of other wireless

communication technologies, such as 5G cellular networks. Additionally, new methods for

calculating the probability of successful delivery could be proposed, advancing the fitness

function, link quality, and link efficiency.

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

34

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

Funding Statement

The authors thank the Pontifícia Universidade Católica de Minas Gerais (PUC-Minas) and

Fundação de Amparo à Pesquisa de Minas Gerais (FAPEMIG - Grant APQ 04034-23) and

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES Grant

88887.340314/2019-00 and 88887.151907/2017-00 ).

References

[1] I. Ahmad, R. M. Noor, Z. Ahmed, N. Akram, and F. P. García Márquez, "A cooperative

heterogeneous vehicular clustering framework for efficiency improvement," Frontiers of

Information Technology & Electronic Engineering, vol. 22, pp. 1247-1259, 2021.

[2] A. A. Hussein and D. A. Mahmood, "Connectivity Analysis in Vehicular Ad-hoc Network

based on VDTN," Journal of Communications Software and Systems, vol. 19, pp. 147-

157, 2023.

[3] M. M. Islam, A. A. R. Newaz, L. Song, B. Lartey, S. C. Lin, W. Fan, et al., "Connected

autonomous vehicles: State of practice," Applied Stochastic Models in Business and

Industry, 2023.

[4] A. Bello Tambawal, R. Md Noor, R. Salleh, C. Chembe, and M. Oche, "Enhanced weight-

based clustering algorithm to provide reliable delivery for VANET safety applications,"

PloS one, vol. 14, p. e0214664, 2019.

[5] Dhinesh Kumar R., Suresh Chavhan, "Shift to 6G: Exploration on trends, vision,

requirements, technologies, research, and standardization efforts", Sustainable Energy

Technologies and Assessments, vol.54, pp.102666, 2022.

[6] Xin Song, Runfeng Zhang, Yu Wang, Yao Yu, Siyang Xu, "Incentive Mechanism Design

for Two-Layer Mobile Data Offloading Networks: A Contract Theory Approach", Ad Hoc

Networks, pp.103154, 2023.

[7] A. M. De Souza, G. Maia, T. Braun, and L. A. Villas, "An interest-based approach for

reducing network contentions in vehicular transportation systems," Sensors, vol. 19, p.

2325, 2019.

[8] H. Zhu, J. Liu, L. Jin, and G. Zhang, "Intersection-Based Unicast Routing Using Ant

Colony Optimization in Software-Defined Vehicular Networks," Electronics, vol. 12, p.

1620, 2023.

[9] N. u. Sehar, O. Khalid, I. A. Khan, F. Rehman, M. A. Fayyaz, A. R. Ansari, et al.,

"Blockchain enabled data security in vehicular networks," Scientific Reports, vol. 13, p.

4412, 2023.

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

35

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

[10] S.-U. Hassan, M. Shabbir, S. Iqbal, A. Said, F. Kamiran, R. Nawaz, et al., "Leveraging

deep learning and SNA approaches for smart city policing in the developing world,"

International Journal of Information Management, vol. 56, p. 102045, 2021.

[11] H. Liu, Z. Guo, and Z. Zhang, "A hybrid multi-level optimisation framework for

integrated production scheduling and vehicle routing with flexible departure time,"

International Journal of Production Research, vol. 59, pp. 6615-6632, 2021.

[12] B. Alaya and L. Sellami, "Towards the Design of an Efficient and Secure System based

on the Software-Defined Network Paradigm for Vehicular Networks," IEEE Access,

2023.

[13] L. Nkenyereye, R. P. Naik, J.-W. Jang, and W.-Y. Chung, "Software-Defined Small Cell-

Linked Vehicular Networks: Architecture and Evaluation," Electronics, vol. 12, p. 304,

2023.

[14] M. Bonanni, "Distributed Services Management over Dynamic Networks of Things:

towards a Unified SDN oriented Design," 2023.

[15] O. Ajibuwa, B. Hamdaoui, and A. A. Yavuz, "A Survey on AI/ML-Driven Intrusion and

Misbehavior Detection in Networked Autonomous Systems: Techniques, Challenges and

Opportunities," arXiv preprint arXiv:2305.05040, 2023.

[16] B. S. Bari, K. Yelamarthi, and S. Ghafoor, "Intrusion Detection in Vehicle Controller Area

Network (CAN) Bus Using Machine Learning: A Comparative Performance Study,"

Sensors, vol. 23, p. 3610, 2023.

[17] Yu, Tzu-Yang, et al. "HetCast: Cooperative data delivery on cellular and road side

network." IEEE 28th Annual International Symposium on Personal, Indoor, and Mobile

Radio Communications (PIMRC). IEEE, 2017.

[18] C. M. Huang, M. S. Chiang, D. T. Dao, W. L. Su, S. Z. Xu and H. Zhou, "V2V data

offloading for cellular network based on the software defined network (SDN) inside

mobile edge computing (MEC) architecture", IEEE Access, vol. 6, pp. 17741-17755,

2018.

[19] P. Si, Y. He, H. Yao, R. Yang and Y. Zhang, "DaVe: Offloading delay-tolerant data traffic

to connected vehicle networks", IEEE Trans. Veh. Technol., vol. 65, no. 6, pp. 3941-3953,

Jun. 2016.

[20] P. A. Apostolopoulos, G. Fragkos, E. E. Tsiropoulou and S. Papavassiliou, "Data

offloading in UAV-assisted multi-access edge computing systems under resource

uncertainty", IEEE Trans. Mobile Comput., Mar. 2021.

[21] J. Xue and Y. An, "Joint Task offloading and resource allocation for multi-task multi-

server NOMA-MEC networks", IEEE Access, vol. 9, pp. 16152-16163, 2021.

[22] Y.-R. Chen, A. Rezapour, W.-G. Tzeng, and S.-C. Tsai, "RL-routing: An SDN routing

algorithm based on deep reinforcement learning," IEEE Transactions on Network Science

and Engineering, vol. 7, pp. 3185-3199, 2020.

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

36

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

[23] C. M. Huang, S. Y. Lin and Z. Y. Wu, "The k-hop-limited V2V2I VANET data offloading

using the mobile edge computing (MEC) mechanism", Veh. Commun., vol. 26, Dec. 2020.

[24] Huang, Chung-Ming, and Chi-Feng Lai. "The delay-constrained and network-situation-

aware V2V2I VANET data offloading based on the multi-access edge computing (MEC)

architecture." IEEE Open Journal of Vehicular Technology 1, 331-347, 2020.

[25] G. Husnain, S. Anwar, and F. Shahzad, "An Enhanced AI-Enabled Routing Optimization

Algorithm for Internet of Vehicles (IoV)," Wireless Personal Communications, pp. 1-21,

2023.

[26] LIN, Shih-Yang; HUANG, Chung-Ming; WU, Tzu-Yu. “Multi-access edge computing-

based vehicle-vehicle-RSU data offloading over the multi-RSU-overlapped

environment”. IEEE Open Journal of Intelligent Transportation Systems, v. 3, p. 7-32,

2022.

[27] Y. Choi and Y. Lim, "Deep Reinforcement Learning for Edge Caching with Mobility

Prediction in Vehicular Networks," Sensors, vol. 23, p. 1732, 2023.

[28] S. Bharany, S. Sharma, N. Alsharabi, E. Tag Eldin, and N. A. Ghamry, "Energy-efficient

clustering protocol for underwater wireless sensor networks using optimized glowworm

swarm optimization," Frontiers in Marine Science, vol. 10, p. 1117787, 2023.

[29] LOUSADA, Efrem EO et al. “Using Complex Networks Metrics to Mitigate the

Broadcast Storm Problem”. IEEE Symposium on Computers and Communications

(ISCC). IEEE, p. 1-6, 2019.

[30] Lima, Sucupira, et al. "E-probt: a new approach to mitigate the broadcast storm problem

in vanets." Proceedings of the 31st Annual ACM Symposium on Applied Computing.

2016.

[31] Karp, Brad, and Hsiang-Tsung Kung. "GPSR: Greedy perimeter stateless routing for

wireless networks." Proceedings of the 6th annual international conference on Mobile

computing and networking. 2000.

[32] Menouar, Hamid, Massimiliano Lenardi, and Fethi Filali. "Movement prediction-based

routing (MOPR) concept for position-based routing in vehicular networks." 2007 IEEE

66th vehicular technology conference. IEEE, 2007.

[33] Jerbi, Moez, et al. "GyTAR: improved greedy traffic aware routing protocol for vehicular

ad hoc networks in city environments." Proceedings of the 3rd international workshop on

Vehicular ad hoc networks. 2006.

[34] Perkins, Charles, Elizabeth Belding-Royer, and Samir Das. "RFC3561: Ad hoc on-

demand distance vector (AODV) routing." (2003).

[35] Y.-R. Chen, A. Rezapour, W.-G. Tzeng, and S.-C. Tsai, "RL-routing: An SDN routing

algorithm based on deep reinforcement learning," IEEE Transactions on Network Science

and Engineering, vol. 7, pp. 3185-3199, 2020.

[36] S. Subramani and M. Selvi, "Deep Learning based IDS for Secured Routing in Wireless

Sensor Networks using Fuzzy Genetic Approach," 2022.

Revista Gestão e Secretariado (GeSec), São Paulo, SP, v. XX, n. X, 202X, p. XX-XX.

An approach for offloading with multi-hop considerations in an RSU

signal overlay setting

37

Revista de Gestão e Secretariado – GeSec, V. 15, N. 4, P.01-37, 2024

São José dos Pinhais, Paraná, Brasil.

[37] G. Arya, A. Bagwari, and D. S. Chauhan, "Performance analysis of deep learning-based

routing protocol for an efficient data transmission in 5G WSN communication," IEEE

Access, vol. 10, pp. 9340-9356, 2022.

[38] JACOBSON, Van. “Congestion avoidance and control”. ACM SIGCOMM computer

communication review, v. 18, n. 4, p. 314-329, 1988.

[39] LIMA, Daniel Sucupira et al. “ProbT: A temporal probabilistic protocol to mitigate the

broadcast storm problem in VANETs”. 2015 international conference on information

networking (ICOIN). p. 7-12. IEEE, 2015.

Received: 03.22.2024

Accepted: 04.12.2024