Edge computing for Vehicle to Everything: a short review

Version Changes

Revised. Amendments from Version 3

The updated version incorporates an enhancement in the form of solidified support for Edge Computing within the ecosystem of Vehicle-to-Everything (V2X) technologies. This version includes an important addition of the European Telecommunications Standards Institute (ETSI) standard specifically designed to govern Edge Computing in V2X systems. Furthermore, this version introduces a clear differentiation between Cellular V2X (C-V2X) and Dedicated Short-Range Communication (DSRC) technologies, acknowledging their distinctive characteristics and functionalities. The version includes relevant references that support the information presented. Lastly, this version also contains a forward-looking stance by probing into the potential future directions of V2X technologies.

Abstract

Vehicle to Everything (V2X) communications and services have sparked considerable interest as a potential component of future Intelligent Transportation Systems. V2X serves to organise communication and interaction between vehicle to vehicle (V2V), vehicle to infrastructure (V2I), vehicle to pedestrians (V2P), and vehicle to networks (V2N). However, having multiple communication channels can generate a vast amount of data for processing and distribution. In addition, V2X services may be subject to performance requirements relating to dynamic handover and low latency communication channels. Good throughput, lower delay, and reliable packet delivery are the core requirements for V2X services.  Edge Computing (EC) may be a feasible option to address the challenge of dynamic handover and low latency to allow V2X information to be transmitted across vehicles. Currently, existing comparative studies do not cover the applicability of EC for V2X. This review explores EC approaches to determine the relevance for V2X communication and services. EC allows devices to carry out part or all of the data processing at the point where data is collected. The emphasis of this review is on several methods identified in the literature for implementing effective EC. We describe each method individually and compare them according to their applicability. The findings of this work indicate that most methods can simulate the EC positioning under predefined scenarios. These include the use of Mobile Edge Computing, Cloudlet, and Fog Computing. However, since most studies are carried out using simulation tools, there is a potential limitation in that crucial data in the search for EC positioning may be overlooked and ignored for bandwidth reduction. The EC approaches considered in this work are limited to the literature on the successful implementation of V2X communication and services. The outcome of this work could considerably help other researchers better characterise EC applicability for V2X communications and services.

Introduction

The automobile industry is changing in various ways, and this provides a chance to address potential transportation-related difficulties. This includes transitioning a traditional independent network to a connected network within and outside the vehicle. ¹ The evolution of drivers’ and passengers’ involvement with vehicles has been evolving both in technology and style. ² Figure 1 depicts the evolution of the interaction between drivers and passengers from 1807 to the present.

Figure 1. Evolution of driver and passenger interactions.

Almost everyone is connected to the Internet, with around six connected devices per person and hundreds of new connections created each second, resulting in billions of connected ecosystems. ³ Furthermore, research has projected that by 2025, connected vehicles will produce over 200 petabytes of data, with at least four terabytes of data generated continuously. This would increase the number of connected vehicles on roads by approximately four hundred million. ⁴ A connected vehicle is one that is equipped with both Internet and wireless LAN connectivity, allowing data to be transmitted between devices both inside and outside of the vehicle. The Internet of Vehicles or Vehicle to Everything (V2X) is a common network for connected vehicles. ^{5 , 6} However, the most challenging problem is efficiently processing and sending enormous data over communication networks.

The difficulty is not just handling data produced by these connected vehicles that are constantly exposed but also maintaining security, deployment, and performance. ^{7 , 8} Therefore, the potential of edge computing (EC) for V2X can play a prominent part. EC is a distributed computer system that carries out computational tasks (such as collecting and analysing data) on a device, particularly a vehicle. In turn, this reduces the transmission of data from the cloud back and forth. ^{9 , 10}

The prospects of V2X and EC has significant potential for enabling a wide range of innovative applications and services for connected and autonomous vehicles. One of the main advantages of EC is its ability to enable real-time decision making for autonomous vehicles. ^{2 , 3} This includes tasks such as detecting and responding to obstacles or hazardous road conditions. By processing sensor data in real-time at the edge of the network, autonomous vehicles can make split-second decisions that enhance safety and efficiency. ⁶ In addition to enhancing safety, EC can also be used to provide personalized infotainment and other services to passengers based on their preferences and location. By analyzing data from a variety of sources, including social media, weather, and traffic patterns, EC can provide passengers with tailored recommendations for entertainment, food, and other services based on their location and preferences. ^{9 , 10} This can enhance the overall passenger experience and help differentiate connected and autonomous vehicles from traditional modes of transportation. This review examines EC, in particular for V2X. We discuss the background of automotive evolution, V2X and EC; prior research on the applicability of EC for V2X; the potential challenges of applying EC to the V2X scenario; and the path for the future.

Automotive evolution, V2X, and EC

V2X communication is a crucial component of current intelligent transportation systems (ITS). For example, V2X provides drivers with information about road hazards that they may overlook. ¹¹ V2X encompasses two main technologies: Cellular V2X (C-V2X) and Dedicated Short-Range Communication (DSRC). C-V2X enables direct communication between vehicles and infrastructure using cellular networks, while DSRC relies on dedicated short-range communication protocols such as IEEE 802.11p. Both C-V2X and DSRC are critical components of current intelligent transportation systems. In addition, V2X allows communication between a vehicle and anything that might impact the environment, including the surrounding infrastructure such as traffic lights (infrastructure) and even smartphones (pedestrian), enabling communication between vehicles and pedestrians holding a smartphone. ¹² With the technology progressing globally, it is just a matter of time until it is widely adopted and deployed. ¹³

EC refers to a technology that allows network-level processing, downstream data for cloud services, and upstream data for IoT service support. ¹⁴ The term “edge” refers to any computer device located in the area between data sources and the cloud. EC is more suitable for applications that require rapid and consistent response times. ¹⁵ V2X is an example, as computing at the edge can reduce data transfer, decreasing reaction times. The European Telecommunications Standards Institute (ETSI) has defined standards for edge computing, which play a crucial role in enabling efficient data processing and communication in the context of V2X services. By adhering to ETSI’s edge computing standard, compatibility and interoperability of edge computing technologies can be ensured, facilitating seamless integration into V2X applications. ¹⁶ For example, when driving, the vehicle captures data via movement, speed, and other sensors, then analyses them to ensure safety and convenience.

For V2X, real-time situational awareness is crucial, particularly on crucial route segments (e.g., an accident is detected by another vehicle on a particular road). Additionally, a backend server will have to provide high-definition local maps. Leveraging local maps and situational awareness is not just about providing data about road traffic conditions. It should also be extended to occurrences where local data must be aggregated in real-time and distributed to drivers on the road through road side units (RSUs). Road users may build and maintain real-time situational awareness using broadcast information from neighbour vehicles as an alternative to EC. Therefore, EC deployment enables shifting such activities to the network edge by combining data from many sources and efficiently broadcasting a huge amount of data to many drivers locally.

EC connects the processing and interpretation of inputs in the direct range to the end devices. The edge is a point of contact between vehicles and the cloud. The edge server has computing and storage means placed close to vehicles. ¹⁷ The services furnished by edge computing, conforming to ETSI standards, produce a substantial level of quality of service to the end user (edge node). The current applications of in-vehicle networks should be supported by efficient communication and computational application. Edge servers analyse and store the data acquired by the sensors in vehicles. Low-latency communication is made possible by these services, which also serve to promote awareness. ^{18 , 19}

Software Defined Networking (SDN) is characterised by the independence of the data plane and control plane. The connection between the data plane and control plane causes packet flow to be delayed, which increases both the cost and the time required for transmission. ²⁰ Thus, this has led the SDN to be concerned about reducing the amount of communication between the data plane and the control plane. The controller can use the real-time central global information provided by the network infrastructure to make informed management decisions. Through the use of a control plane, the controller has access to a network view. ^{21 , 22} It communicates with forwarding devices and offers network control applications with an integrated development environment. As a result, it helps the dynamic adoption and administration of networks without impacting the operations of existing network infrastructures. It helps in the development of new applications and efficient use of resources. The configuration of the SDN control plane enables the network to adapt to changes in network dynamics caused by unforeseen situations, such as traffic accidents or lane diversions. The SDN works in coordination with edge computing and supports the transportation systems’ ability to engage with the uncertainty of a varying and large number of vehicles, massive data volumes, and frequent disconnections to enable immediate vehicle connectivity and thus enhance the safety of the driver. ^{23 – 25}

The architecture of EC is composed of three layers: cloud layer, edge cloud layer, and the vehicle network. A number of essential advantages of the cloud layer encompass batch processing, predictive analytics, analytical modelling, caching, high scalability, and data processing of detailed information, which are all far beyond the computing capabilities of the edge devices. ²⁶ The cloud can process large volumes of data in a short time. Cloud infrastructure includes storage and processing. The data collected among many edge nodes will be transmitted to the cloud layer, entirely stored for long-term analysis. Since the data may be utilized over a more extended period, it would not be necessary to do computations in real-time. Components for computing and analysis conduct complicated computations more quickly. Unlike latency-sensitive services, computational processes transmitted suffer from delay. ²⁷

On the other hand, the edge cloud layer links the vehicle network and the cloud layer. In order to do this, the vehicles are equipped with communication hardware that makes use of wireless communication protocols such as wireless LAN (IEEE 802.11n), dedicated short-range communication (IEEE 802.11p), cellular networks (4G/5G), and LoRa to communicate with one another and with their surroundings. ^{28 , 29} The purpose would be to provide reliable communication, emergency services, situational awareness, content identification, storage, processing, and increasing efficiency because it is situated close to vehicles and is used for reliable information. The edge cloud layer is beneficial for applications that need a faster response with very low delay, such as environment detection (flash flood), health recognition (drivers’ health condition), and driving behaviour identification. ^{30 , 31}

Vehicles are expected to conduct more services. The vehicle network layer includes a collection of topologically nearby vehicles and shares computational and storage capabilities over a wireless network. ³² The vehicle network abstracts information from an on-board unit (OBU), a global positioning system (GPS), embedded sensors, cameras, lidar, radar, and other devices. The information obtained may be transmitted to the edge cloud layer for storing or utilized as a source for services offered at the application layer. As part of the vehicle network, the vehicle will be equipped with communications, intelligence, storage, and adapting capabilities that will allow it to expect the driver’s expectations. Additionally, the concept will also assist by offering offloading and computational capabilities for vehicles delivering all preferred services at the edge network. ^{33 , 34}

Applicability of EC for V2X

As of recent years, V2X and EC technologies are still in the early stages of development and deployment. However, there have been significant advancements in recent years, and many researchers and industry experts are optimistic about the future of these technologies. ^{28 , 31} In terms of V2X, many automakers and technology companies are investing in the development of connected and autonomous vehicles. In addition, several standards organizations, such as the IEEE and the 3GPP, are working to establish common standards and protocols for V2X communication. These standards will help ensure interoperability between different vehicles and systems and enable the development of a robust V2X ecosystem. ^{29 , 30} As for EC, there has been a rapid expansion of edge devices and infrastructure, driven by the increasing demand for low-latency and high-bandwidth applications such as autonomous driving and smart cities. Many cloud providers, such as Amazon Web Services and Microsoft Azure, are also expanding their edge computing capabilities to meet the growing demand for edge services. Moreover, there are several open-source projects and industry initiatives, such as the Open Edge Computing Initiative, aimed at promoting the development and adoption of EC technologies.

EC applies to a wide range of uses, from sensor applications (e.g., predictive vehicle maintenance) to the end-user experience (e.g., collision prevention warning). EC has been discussed previously from the perspective of V2X communication applicability. In 2022, Pack proposed a lightweight vehicular edge computing framework to address the security and privacy challenges in V2X communications. The authors introduced the LS-VEC framework, which leverages directed acyclic graphs (DAGs) to record transactions for edge resource allocation and micro-transactions for pricing VEC resources. ³⁵ Additionally, an auction theory-based game-theoretic approach is proposed for efficient allocation and pricing of edge resources supporting computation offloading. The LS-VEC framework presents a promising solution to enhance the security and privacy of V2X services while maintaining low latency and efficient resource allocation. In the same year, Khan et al. introduced a modular infrastructure for V2X, presenting “B-UV2X,” a blockchain hyperledger fabric-enabled distributed permissioned network-based consortium structure. The participating vehicle nodes are interconnected within the smart city chain, exchanging various information such as movement data, which is securely logged on the blockchain-enabled immutable ledger. ³⁶ The implementation and evaluation of B-UV2X demonstrate its potential to automate IoV transactions with doppler spread, presenting a promising solution for V2X communication.

Vladyko et al. addressed the challenge of reliable and secure information exchange in real-time within the vehicle-to-everything (V2X) concept. To achieve this, the authors propose a network architecture leveraging blockchain technology and mobile edge computing (MEC). The authors present a formalized mathematical model considering the interconnection of objects and V2X information channels. An energy-efficient algorithm for traffic offloading to the MEC server is also introduced. ³⁷ Through the implementation and evaluation of the proposed approach using blockchain and MEC in the developed system, the proposed approach demonstrates advantages and disadvantages of the implementation, highlighting the potential of blockchain and edge computing in ensuring reliable and secure V2X communications. Awais et al. explored the potential of mobile edge computing (MEC) and blockchain technology. The proposed architecture adopts a hierarchical computing approach with cloud computing (CC), edge computing (EC), and fog computing (FC) nodes to distribute resources and databases based on computing service requirements. A blockchain-based transaction service is integrated to enhance reliability. ³⁸ The approach is experimentally evaluated in two use cases: plug-in electric vehicles in smart grid scenarios and massive IoT data services for autonomous cars. The results demonstrate reduced network consumption, improved security features, and lower computational load for computed log storage compared to a centralized network system.

Al-Bhakri et al. developed a simulation models for automotive networks based on mobile edge computing (MEC) and software-defined networking (SDN) technologies. The architecture of the automotive network is to enhance the management of network resources, applications, and users. There were two experiment conducted, the first experiment demonstrates the positive effect of edge and fog computing on the network core load, while the second experiment shows the positive effect of SDN in load balancing between base stations (BS) and roadside units (RSU). ³⁹ Xu et al. proposed a computation offloading method called V2X-COM that utilizes vehicle-to-everything (V2X) communication to improve smart vehicle applications in the internet of vehicles (IoV). The method employs vehicle-to-infrastructure communication (V2I) and vehicle-to-vehicle communication (V2V) to determine feasible routes for offloading computation tasks to edge nodes (ENs). The study uses non-dominated sorting genetic algorithm III (NSGA-III), simple additive weighting (SAW), and multiple criteria decision making (MCDM) to generate balanced and optimal offloading strategies, which are experimentally evaluated for validation. ⁴⁰

Moubayed et al. described an Optimum V2X Service Placement (OVSP) as a binary integer issue in a linear edge context. ⁴¹ The authors approached this problem using a low-complexity greedy heuristic technique (G-VSPA). Extensive simulations showed that the OVSP model provides satisfactory results when sensitive services to delays are on the edge and tolerant services to delays are at the core of the process. Furthermore, the proposed algorithm provides near-optimal performance with minimal complexity. In the same year, Shaer et al. addressed the efficient deployment of V2X essential services, including various V2X applications in the EC environment. ⁴² The authors devised an optimisation method for minimising E2E latency in multi-component V2V systems under different traffic situations. The findings indicate that the methodology guarantees an adequate level of service and surpasses solutions developed in earlier studies using realistic scenarios.

Lee et al. described an EC approach for minimising trip time at interconnected junctions. ⁴³ The authors suggested a paradigm in which each road side unit (RSU) determines junction scheduling while the vehicles select their travel trajectory through dynamic control. Based on simulation results for optimum scheduling of linked junctions, the proposed framework significantly reduced overall travel time by up to 14.3%. Grammarikos and Cottis investigated the benefits of mobile edge computing (MEC) adopting V2X services linked to traffic efficiency and road safety. ⁴⁴ A simulation model that represented a long-term evolution (LTE) system with basic MEC capabilities, such as packet routing, was investigated in this work to evaluate the applicability of their findings. The presented approach evaluated the packet delivery ratio and packet loss for applications, such as telemetry and emergency message delivery, respectively. While LTE can transmit traffic data to vehicles in a short amount of time, the simulation results revealed that severe congestion in the backhaul and core networks could result in unexpected packet losses, which could be prevented by the processing capabilities of a MEC server.

In addition, Napolitano et al. proposed a fully compatible design and implementation of a vulnerable road users (VRU) warning system, as well as an experimental assessment of the system using MEC- and cloud-based architectures. ⁴⁵ The authors developed a strategy that would enable road users to communicate information regarding the existence of neighbouring entities in the event of a difficult circumstance (e.g., road accident). This is accomplished by using an architecture that consists of a user-facing Android application and a MEC-based application [cooperative awareness messages (CAM)]. The E2E latency demonstrated a substantial result when visualising the entities engaged between the VRUs application and the CAM server using a preliminary performance measurement. Additionally, Emara et al. focused on the case of VRU, examining the safe interaction of vehicles with road users such as motorcyclists and pedestrians. ⁴⁶ The authors aimed to describe latency improvements using MEC systems through periodic CAM. Extensive simulation results indicated that installing MEC infrastructure may substantially decrease the communication latency. Additionally, Sabella et al. suggested a hierarchical MEC architecture for adaptive video streaming in V2X applications. ⁴⁷ The authors described the acquisition of real-time channel data by local agents stationed at the evolved NodeB (eNB). This information is then communicated to a MEC platform, which automatically changes the video stream’s quality to match the channel’s conditions. Within a virtualized network context, the authors tested and evaluated a conceptual demonstration of radio-aware video optimization. The results demonstrated that the proposed architecture enhanced the user experience by boosting downlink and uplink speeds and reducing delay.

Bissmeyer et al. introduced a network framework that ensures V2X information and data exchange in a MEC-based multi-access technology environment. ⁴⁸ The authors designed a framework for the integrity of the message, sender authorisation and authentication, and replay detection. This approach is achieved through digital signatures, an authorisation certificate, and public and private key infrastructure. MEC offers local processing capabilities for the exchange of event-driven V2X encrypted messages within the framework. In addition, Balid et al. demonstrated MEC traffic management methods for real-time traffic monitoring. ⁴⁹ The authors developed and deployed a cost-effective wireless sensor traffic monitoring system for highway and roadside traffic. The sensor achieved an acceptable level of accuracy in terms of detection, speed prediction, and vehicle categorisation. Table 1 provides a comprehensive summary of the previous studies, presenting key information on each research work related to V2X and edge computing (EC).

Table 1. Comprehensive summary of previous studies on V2X and edge computing (EC).

Authors	Title	Technology/Concept	Key findings
35	LS-VEC: Lightweight Vehicular Edge Computing Framework	Directed Acyclic Graphs (DAGs), Auction Theory-based Game-theoretic Approach	Enhanced security & privacy in V2X communications with efficient resource allocation and pricing
36	B-UV2X: Blockchain-enabled V2X Consortium Structure	Blockchain (Hyperledger Fabric), V2X	Secure and automated V2X transactions, potential for doppler spread in V2X communication
37	V2X with Blockchain & Mobile Edge Computing (MEC)	MEC, Blockchain	Reliable & secure V2X information exchange, energy-efficient traffic offloading to MEC server
38	Hierarchical Computing with MEC & Blockchain	MEC, Blockchain	Reduced network consumption, improved security features, and lower computational load compared to centralized networks
39	Automotive Network with MEC & SDN	MEC, SDN	Positive effects of edge and fog computing on network core load, SDN load balancing between base stations and roadside units
40	V2X-COM: Computation Offloading with V2X Communication	V2X Communication, NSGA-III, SAW, MCDM	Offloading computation tasks to ENs using V2I and V2V communication, optimal offloading strategies generated using NSGA-III, SAW, and MCDM
41	Optimum V2X Service Placement with Greedy Heuristic	Binary Integer Issue in a Linear Edge Context	Satisfactory results for OVSP model with near-optimal performance and minimal complexity
42	Minimizing E2E Latency in Multi-component V2V Systems	Optimization Method	Adequate level of service with reduced E2E latency
43	EC Approach for Minimizing Trip Time at Interconnected Junctions	Dynamic Control, RSU Scheduling	Significant reduction in overall travel time by up to 14.3%
44	MEC Adopting V2X Services for Traffic Efficiency	LTE System with Basic MEC Capabilities	Evaluation of packet delivery ratio and packet loss for V2X applications
45	VRU Warning System using MEC & Cloud-based Architectures	MEC, Cloud-based Architecture	Substantial reduction in E2E latency in VRU warning system
46	Safe Interaction of Vehicles with VRUs using MEC	MEC	Substantial decrease in communication latency in VRU interaction
47	Hierarchical MEC Architecture for Adaptive Video Streaming in V2X	Hierarchical MEC Architecture	Enhanced user experience by boosting downlink and uplink speeds, reducing delay.
48	V2X Information Exchange in MEC-based Multi-access Technology Environment	Digital Signatures, Public & Private Key Infrastructure	Ensuring message integrity, sender authorization & authentication, and replay detection through digital signatures and key infrastructure within MEC framework.
49	MEC Traffic Management for Real-time Traffic Monitoring	MEC, Wireless Sensor Traffic Monitoring	Cost-effective wireless sensor traffic monitoring system with acceptable accuracy for detection, speed prediction, and vehicle categorization.

Challenges of V2X and EC

Security

At the edge of a network, privacy and security protection are critical services to provide. ^{50 , 51} If the vehicle is equipped with IoT, it can collect sensitive data from sense data. ⁵² Several ITS implementations would need drivers to grant access to sensitive, confidential data to untrusted vehicles attempting to join as edges in the context of smart cities. ⁵³ Together with data segregation techniques, effective trust management systems may considerably increase edge security. ⁵⁴ According to El-Sayed & Chaqfeh, ⁵⁵ although minimal research has been conducted on assuring secure collaboration in an EC scenario, the study does not explicitly address V2X issues.

Deployments

The positioning of edge devices in an urban environment is based on static and dynamic features. ⁵⁶ Edge nodes may need MEC servers with fixed RSUs or unmanned aerial vehicles (UAV). ⁵⁷ Many possible ITS applications may be facilitated by autonomous UAVs, improving traffic safety and transportation quality of life. ⁵⁸ Nevertheless, specific issues must be addressed, such as limited energy, processing ability, and signal transmission range. ⁵⁹ Given the technological developments such as sensor-based street lights or smart toll booths over the past few decades, the limitations on UAV usage will likely be overcome eventually.

Performances

Each second counts when you’re behind the wheel of a vehicle. As a result, vehicles would continuously upload the data collected by their local sensors to the closest edge device. ^{60 , 61} Hence, energy and power consumption at the edge should be considered to avoid service disruptions and quality of service (QoS) loss. ^{62 , 63} Furthermore, various situations need substantial QoS improvement to cope with occasional high traffic loads like severe traffic congestion, unpredicted weather conditions, or unexpected road construction works. ⁶⁴ Therefore, further research is necessary to enhance and manage QoS in the V2X context considering a heterogeneous edge-based environment.

Conclusions and future direction

EC adoption is growing in the automotive industry, and ITS, particularly V2X, will certainly change various economic sectors and significantly influence our everyday lives. Despite this, multiple different challenges are limiting its wide implementation. The increasing number of sensors in connected vehicles and roads creates a large data processing and storage issue. This requires new service platforms with strong processing, reliable storage, and real-time communication. EC is indeed a promising way to decrease latency and bring data closer to vehicles and resources. In the future, we will work on a comprehensive middleware solution for V2X communication. In many V2X scenarios, data transmitted between users and network infrastructure is localised and does not need remote access to centralised data centres. Using EC may substantially improve the performance of supporting various applications of V2X. The availability of network resources, storage, and computation near the network edge make EC an ideal option for V2X delay-sensitive applications.

To move towards a promising future, researchers and industry stakeholders need to focus on several key directions. Firstly, substantial efforts should be dedicated to developing robust security mechanisms for V2X communication at the edge, including privacy protection, effective trust management systems, and secure collaboration protocols. Secondly, innovative solutions must be explored to optimize the positioning and deployment of edge devices in urban environments, considering static and dynamic features and overcoming limitations related to energy, processing ability, and signal transmission range. Thirdly, there is a critical need to invest in research and development to enhance the performance of V2X applications in heterogeneous edge-based environments, focusing on reducing energy and power consumption while ensuring reliable quality of service, especially in high traffic load scenarios. Additionally, the creation of comprehensive middleware solutions tailored to V2X communication can significantly contribute to addressing data processing and storage challenges and streamline communication between users and network infrastructure. By adopting these future directions, V2X and EC can unlock their full potential, revolutionizing various economic sectors and significantly enhancing the safety and convenience of our everyday lives on the road.

Data availability

No data is associated with this article.

Funding Statement

The authors would like to acknowledge Multimedia University for the approved IRFUND MMUI/210026 which makes this important research viable and effective. The authors gratefully acknowledge the use of service and facilities of the Connected Car Services Research Group, Centre of Intelligent Cloud Computing at Multimedia University. The authors would also like to thank the anonymous reviewers for their constructive comments.