Link Lifetime Prediction in Vehicular Ad Hoc Networks: Comparison
View Latest Version
Please note this is a comparison between Version 2 by Beatrix Zheng and Version 1 by Tuan Anh Nguyen.

In urban mobility, Vehicular Ad Hoc Networks (VANETs) provide a variety of intelligent applications. By enhancing automobile traffic management, these technologies enable advancements in safety and help decrease the frequency of accidents. The transportation system can now follow the development and growth of cities without sacrificing the quality and organisation of its services thanks to safety apps that include collision alerts, real-time traffic information, and safe driving applications, among others. Applications can occasionally demand a lot of computing power, making their processing impractical for cars with limited onboard processing capacity. Offloading of computation is encouraged by such a restriction. However, because vehicle mobility operations are dynamic, communication times (also known as link lifetimes) between nodes are frequently short. VANET applications and processes are impacted by such communication delays (e.g., the offloading decision when using the Computational Offloading technique). Making an accurate prediction of the link lifespan between vehicles is therefore challenging.

  • vehicle networks (VANETs)
  • computational offloading
  • machine learning
  • features
  • V2V communication
  • IEEE 802.11p
  • SUMO
  • NS-3

1. Introduction

According to [1], the number of vehicles was calculated at 1.2 billion globally in 2014 and predicted to reach 2.0 billion in 2035. Annually, vehicles threaten the lives of about 1.3 million people around the world. It is predicted that in 2030, road accidents will be the fifth highest cause of death in the world [2]. The majority of accidents happen in urban environments and are caused by improper traffic coordination [3]. As a result, there is a need to design systems that improve traffic efficiency and safety while also offering consumers a variety of services and traffic-related information. This area has been called ITS (Intelligent Transportation Systems) [4]. A key reason for implementing ITS is the increase in the number of cars capable of connecting to massive vehicular networks and sharing data [5]. Vehicle Ad Hoc Networks (VANETs) are crucial to the deployment of ITS because they provide a variety of functions, including the supply of real-time traffic information. For example, such data may be utilised to optimise traffic flow and reduce traffic congestion [6]. Ad hoc networks, wireless LANs, mobile technologies, and sensor networks are all integrated into VANETs to enable improved intelligent communications between cars, roadside sensors, and infrastructure. This makes VANETs a versatile and complex technology [7]. Different communication channels supported by VANET include Vehicle to Vehicle (V2V), Vehicle to Infrastructure (V2I), and Hybrid channels [8]. New sophisticated applications have evolved as a result of the VANETs industry’s ongoing growth, including real-time surveillance, augmented reality, streaming video in cars, autonomous driving, and more [9]. Such applications pose a challenge to a vehicle’s constrained resources since some of them need processing power and are latency-sensitive [10].
Utilizing computational offloading techniques, which shift complete workloads to other devices to make better use of the resources available, is one approach to increase the performance of applications in VANETs. The offloading procedure in VANETs uses idle resources from neighbouring cars to conduct tasks or programmes on those devices [11]. Due to frequent route failures brought on by the high mobility of vehicles, delay and cost in computational offloading are seldom managed adequately [12]. Data packet transfers between the origin and destination vehicles frequently encounter interruptions due to the communication route’s limited lifespan, necessitating re-transmissions [13]. Offloading is strongly dependent on techniques that estimate the link lifetime because of issues with communication interruption; this is important in a computational offloading choice process in VANET [14].
Recent research views the ability to estimate the link lifespan between two vehicles as an enabler for enhancing VANET communication quality [15]. The link lifespan is not accurately predicted by the available solutions. The link lifespan prediction method is now mostly carried out by using computations and various equations, which has three primary issues or limitations. (i) Low lifetime inference accuracy can result in high error rates, or a larger difference between the predicted lifetime value and the actual value. (ii) The prediction calculus theorems only take into account a small number of features, making it challenging to incorporate new features into the prediction solution. The term “feature” in this restudyearch refers to the contextual traffic data that is gathered and used as labeled input data or variables in the prediction process. Utilizing few qualities contrasts with the development of vehicle networks, which, because of their numerous sensors and shared information, enable a more notable emergence of new features [12,16] [12][16]. (iii) A small selection of analysed scenarios are either in the experiments or in the theories put out in the literature. In the data collecting stage, the majority of studies concentrates on just one kind of scenario. As time goes on, the unique traits and quirks of each vehicle’s communication environment are eventually neglected, distorting the prediction process and yielding predictions of values that are inconsistent with each context.
The lifespan prediction procedure is generally limited by machine learning (ML) techniques. Unrelated to offloading, it is noteworthy that the authors in [17] employed an ML method called Adaboost to predict the link lifespan. Other studies employ mathematical equation-based systems with a maximum of three elements that take into account the position, speed, acceleration, or direction of the vehicle. Few studies that concentrate on estimating link lifetimes make use of that data in computational offloading; examples are [14,18,19,20][14][18][19][20]; nevertheless, the settings under consideration are straightforward. Finally, apart from [21], which incorporates an urban environment into its prediction scheme, various prediction environments are not addressed, with related work concentrating primarily on the highway environment.

2. Machine Learning Techniques Related to VANET

Machine learning techniques are used in different ways to solve problems related to VANET. Several works emphasize using ML techniques in the decision-making optimization process in computational offloading algorithms in task offloading. The authors in [18] group vehicles to form clusters to increase the reliability of the link between nodes for efficient offloading of tasks. Next, the authors formulate an optimization problem to minimize energy consumption and task offloading latency. In [22], the authors consider collaborative computational offloading in a dynamic edge-cloud network and formulate an optimization problem in the offloading decision process. They use approaches based on deep reinforcement learning to solve the task optimization problem, thus reducing energy consumption and communication delay. Some works use machine learning approaches to predict mobility in Vanet, as in [23], which proposes a routing model based on a hybrid metaheuristic algorithm combined with ensemble learning. The authors calculate model execution using various machine learning techniques, including SVM, Naive Bayes, ANN, and decision tree. In [24], the authors implement a network traffic prediction model considering the parameters that can lead to road traffic. The proposed model integrates a random forest and network traffic prediction algorithm to simultaneously predict the network traffic flow based on road and network traffic.
As seen, the use of machine learning is quite common in the development of vehicular networks, with applications in numerous fields. However, using learning techniques in the link lifetime part is still challenging, with few works in this field. The authors do not seek to extend their applications or processes for link lifetime prediction.
WThe researchers analyzed that machine learning approaches in vehicular networks are used in some fields, but there is still a shortage of work in the process of lifetime prediction. With this, wethe researchers seek to raise the current state of prediction work concerning characteristics, processes, and use cases, among others. WeThe researchers compare some main works in the literature which implement link lifetime prediction processes with our worktheir own research, highlighting characteristics that differentiate them. Table 1 summarizes the main points of analysis.
Table 1. Comparison with related work.
Work Prediction Features Scenario Does It Use ML

in the Prediction?		 Does It Use Offloading? Contribution
 [21] Distance, Direction,

and Speed		 Urban No No A new protocol called MPBRP (Mobility

Prediction Based Routing Protocol) seeks

to predict the mobility of nodes through

positions and angles to predict the best route.
[25] Relative Speed

and Position		 Highway No No A method predicts the TVE between

two moving vehicles based on their

relative speed.
[26] Position, Speed,

and Acceleration		 Highway No No Implementation of two protocols. One is to

predict the most stable route, and another is

to predict the time of packet delivery before

sending the data. A prediction scheme of

LLT was developed, aiming to guarantee

the efficiency of protocols.
[14,18][14][18] Position, Speed,

and Angle		 Highway No Yes Task offloading scheme for cellular vehicle

to everything (C-V2X). Performance

evaluation with other offloading schemes

through a road scenario simulation.
[27] Speed, Distance, and

Maximum Transmission

Range		 Highway No No Introduction of a model to use the Residual

Link Lifetime and one-hop latency in VANET.
[20] Speed and Distance Highway No Yes A offloading method in V2V communication

by implementing an SDN (Software Defined

Network) within a MEC (Mobile Edge

Computing) architecture, abbreviated as the

SDNi-MEC server.
[19] Position, Speed,

and Acceleration		 Highway No Yes A scheme to improve offloading performance

in multi-interface networks, through

a greedy choice algorithm called GCF.
[17] Link Lifetime,

SLS, Function

Similarity, Relative

Mobility, Distance,

and Speed		 Highway Yes No The authors implemented an LLT prediction

method in VANET environments using the

Adaboost Machine Learning algorithm.

They performed a benchmark between

Adaboost and other ML methods.
This woresearkch Distance, Client Angle,

Server Angle, Pseudo-

angle, Speed, Relative

Speed, Relative

Acceleration, and SLS		 Highway

and Urban		 Yes Yes WThe researchers propose a prediction scheme of the Link

Lifetime through the SVR Machine

Learning algorithm, chosen through a

performance evaluation among other

ML methods. WThe eresearchers extend the GCF algorithm

with a SVR-based prediction function.
The high mobility of the vehicular scenario requires that specific features, such as speed, position, and angles, among others, be observed and considered. In [21], the authors propose a new MPBRP (Mobility Prediction Based Routing Protocol) that uses positions and angles to predict the best path.
In [25], the authors address a method that predicts the lifetime of the link between two moving vehicles based on their relative speed. Subsequently, this lifetime prediction algorithm is implemented with the Ad Hoc On-Demand Distance Vector (AODV) protocol, creating a new protocol called AODV with Lifetime Prediction (AODV-LP).
In [26], the authors propose two protocols, one to predict the most stable route and the other to predict the delivery time of packets before sending the data. A link lifetime prediction scheme was developed to guarantee the protocols’ efficiency. The authors consider the acceleration and deceleration of vehicle speed in direct communication between the two vehicles. In addition to acceleration, the speed and positioning of vehicles are features used when calculating the link lifetime. The authors consider a road scenario in the simulations, not performing tests in other scenarios, such as urban ones.
de Souza et al. [19] proposes a scheme to improve the offloading performance of computationally intensive applications while dealing with the mobility challenges of vehicular environments. The results show that the proposed scheme reduces the total download time by up to 54.1% and increases the download success rate by 71.8% compared to other schemes. However, the work does not use an efficient link lifetime prediction algorithm. OuThis r workesesarch proposes using machine learning algorithms to more realistically predict link lifetime and thus increase offloading success rates.
In [17], a link lifetime prediction method is proposed in VANET environments using the machine learning algorithm Adaboost. A benchmark between Adaboost and other machine learning algorithms was performed, but the authors do not include link lifetime prediction solutions based on mathematical formulations in the benchmark. The simulations considered a highway environment, with urban scenarios not being used. Based on the evaluated regression metrics, outhe researchers' results in the highway scenario showed lower error rates than the highway scenario implemented by the authors.
In [14], a task offloading scheme is proposed in vehicular cellular communication for anything. Vehicles are grouped within the cluster, sending tasks to vehicles within the same cluster via V2V communication or with an edge server. To perform the lifetime prediction of the offloading algorithm proposed by the authors, they use a prediction function based on position, speed, and angle features. The authors consider only the highway scenario regarding the environment adopted in the experiments. In [18], the authors extend the scheme to support offloading on VEC (Vehicular Edge Computing) servers and improve the matching algorithm to optimize offloading decisions.
In [27], the authors use link lifetime prediction to identify nearby vehicles or nodes to perform a hop to a destination vehicle since location-based network protocols in VANETs employ the hop strategy to acquire the route to a node of destination. The authors propose an algorithm to calculate the lifetime of the nearby vehicle before making the jump. The authors model the proposal in a highway environment with dense traffic, considering the speed of both vehicles, the distance, and the maximum transmission range as prediction features.
In [20], the authors propose an offloading method in V2V communication by implementing an SDN (Software Defined Network) within an MEC (Mobile Edge Computing) architecture. The vehicles send their context data to the MEC, and the SDN controller calculates information regarding the ideal route between client and server vehicles. To calculate this route, the authors propose a method to calculate the link lifetime in the SDN controller. This method is calculated based on two features: relative speed and distance.
In the offloading process, the studies presented employ restricted lifespan prediction techniques. Only one study used ML-based lifespan prediction approaches; however, the study does not apply this prediction process to offloading in vehicle networks. In a decision-making approach to task offloading in vehicular communication, wthe researchers apply machine learning approaches to the prediction link lifespan. The paperresearch proposes eight aspects to address the challenge of predicting link lifespan in VANETs. Some proposed characteristics, such as pseudo-angle, angle, and SLS, have not been considered in previous publications. Another significant distinction is that wethe researchers evaluate prediction efficiency using an urban scenario, which is rarely used in earlier publications due to its complexity.

References

  1. Moorhead, J.; Nixon, T. Carbon Pricing on the Horizon. Available online: https://www.oebu.ch (accessed on 3 August 2022).
  2. Kelarestaghi, K.B.; Foruhandeh, M.; Heaslip, K.; Gerdes, R. Survey on vehicular ad hoc networks and its access technologies security vulnerabilities and countermeasures. arXiv 2019, arXiv:1903.01541.
  3. Karunathilake, T.; Förster, A. A Survey on Mobile Road Side Units in VANETs. Vehicles 2022, 4, 29.
  4. Zhang, J.; Wang, F.Y.; Wang, K.; Lin, W.H.; Xu, X.; Chen, C. Data-driven intelligent transportation systems: A survey. IEEE Trans. Intell. Transp. Syst. 2011, 12, 1624–1639.
  5. Salem, A.H.; Damaj, I.W.; Mouftah, H.T. Vehicle as a Computational Resource: Optimizing Quality of Experience for connected vehicles in a smart city. Veh. Commun. 2022, 33, 100432.
  6. Raut, C.M.; Devane, S.R. Intelligent transportation system for smartcity using VANET. In Proceedings of the 2017 International Conference on Communication and Signal Processing (ICCSP), Tamilnadu, India, 6–8 April 2017; pp. 1602–1605.
  7. Hamdi, M.M.; Audah, L.; Rashid, S.A.; Alani, S. VANET-based traffic monitoring and incident detection system: A review. Int. J. Electr. Comput. Eng. 2021, 11, 3193–3200.
  8. Hasrouny, H.; Samhat, A.E.; Bassil, C.; Laouiti, A. VANet security challenges and solutions: A survey. Veh. Commun. 2017, 7, 7–20.
  9. Alabbasi, A.; Aggarwal, V. Joint information freshness and completion time optimization for vehicular networks. IEEE Trans. Serv. Comput. 2020, 15, 1118–1129.
  10. He, Y.; Zhai, D.; Huang, F.; Wang, D.; Tang, X.; Zhang, R. Joint task offloading, resource allocation, and security assurance for mobile edge computing-enabled UAV-assisted VANETs. Remote Sens. 2021, 13, 1547.
  11. Li, C.; Wang, S.; Huang, X.; Li, X.; Yu, R.; Zhao, F. Parked vehicular computing for energy-efficient Internet of vehicles: A contract theoretic approach. IEEE Internet Things J. 2018, 6, 6079–6088.
  12. Deng, Z.; Cai, Z.; Liang, M. A multi-hop vanets-assisted offloading strategy in vehicular mobile edge computing. IEEE Access 2020, 8, 53062–53071.
  13. Lee, M.; Atkison, T. Vanet applications: Past, present, and future. Veh. Commun. 2021, 28, 100310.
  14. Bute, M.S.; Fan, P.; Liu, G.; Abbas, F.; Ding, Z. A Collaborative Task Offloading Scheme in Vehicular Edge Computing. In Proceedings of the 2021 IEEE 93rd Vehicular Technology Conference (VTC2021-Spring), Helsinki, Finland, 25–28 April 2021; IEEE: Piscataway, NJ, USA, 2021; pp. 1–5.
  15. Khatri, S.; Vachhani, H.; Shah, S.; Bhatia, J.; Chaturvedi, M.; Tanwar, S.; Kumar, N. Machine learning models and techniques for VANET based traffic management: Implementation issues and challenges. Peer-to-Peer Netw. Appl. 2021, 14, 1778–1805.
  16. Zhang, J.; Ren, M.Y.; Labiod, H. Performance evaluation of link metrics in vehicle networks: A study from the cologne case. In Proceedings of the 6th ACM Symposium on Development and Analysis of Intelligent Vehicular Networks and Applications, Malta, Malta, 13–17 November 2016; pp. 123–130.
  17. Zhang, J.; Ren, M.; Labiod, H.; Khoukhi, L. Link duration prediction in vanets via adaboost. In Proceedings of the GLOBECOM 2017—2017 IEEE Global Communications Conference, Singapore, 4–8 December 2017; IEEE: Piscataway, NJ, USA, 2017; pp. 1–6.
  18. Bute, M.S.; Fan, P.; Liu, G.; Abbas, F.; Ding, Z. A cluster-based cooperative computation offloading scheme for C-V2X networks. Ad Hoc Netw. 2022, 132, 102862.
  19. De Souza, A.B.; Rego, P.A.L.; Rocha, P.H.G.; Carneiro, T.; de Souza, J.N. A task offloading scheme for wave vehicular clouds and 5G mobile edge computing. In Proceedings of the GLOBECOM 2020—2020 IEEE Global Communications Conference, Taipei, Taiwan, 7–11 December 2020; IEEE: Piscataway, NJ, USA, 2020; pp. 1–6.
  20. Huang, C.M.; Chiang, M.S.; Dao, D.T.; Su, W.L.; Xu, S.; Zhou, H. V2V data offloading for cellular network based on the software defined network (SDN) inside mobile edge computing (MEC) architecture. IEEE Access 2018, 6, 17741–17755.
  21. Ye, M.; Guan, L.; Quddus, M. Mpbrp-mobility prediction based routing protocol in vanets. In Proceedings of the 2019 International Conference on Advanced Communication Technologies and Networking (CommNet), Rabat, Morocco, 12–14 April 2019; IEEE: Piscataway, NJ, USA, 2019; pp. 1–7.
  22. Li, Y.; Xu, S.; Li, D. Deep Reinforcement Learning for Collaborative Computation Offloading on Internet of Vehicles. Wirel. Commun. Mob. Comput. 2021, 2021, 6457099.
  23. Marwah, G.P.K.; Jain, A. A hybrid optimization with ensemble learning to ensure VANET network stability based on performance analysis. Sci. Rep. 2022, 12, 1–20.
  24. Sepasgozar, S.S.; Pierre, S. Network Traffic Prediction Model Considering Road Traffic Parameters Using Artificial Intelligence Methods in VANET. IEEE Access 2022, 10, 8227–8242.
  25. Kulla, E.; Morita, S.; Katayama, K.; Barolli, L. Route Lifetime Prediction Method in VANET by Using AODV Routing Protocol (AODV-LP). In Complex, Intelligent, and Software Intensive Systems; Barolli, L., Javaid, N., Ikeda, M., Takizawa, M., Eds.; Springer International Publishing: Cham, Switzerland, 2019; pp. 3–11.
  26. Nabil, M.; Hajami, A.; Haqiq, A. Predicting the route of the longest lifetime and the data packet delivery time between two vehicles in VANET. Mob. Inf. Syst. 2019, 2019, 2741323.
  27. Pasha, M.; Farooq, M.U.; Ramsha, A.A. An Efficient Scheme for Reliable File Transfer using Residual Link Lifetime Estimation based on Sparse Vehicular Routes. J. Phys. Conf. Ser. 2021, 1964, 042007.
More
© Text is available under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/)
Video Production Service
Feedback