PaperlayerView exact revision on arXiv

Source-linked AI summary

Vehicular Edge Computing and Networking: A Survey

Lei Liu, Chen Chen, Qingqi Pei, Sabita Maharjan, Yan Zhang

arXiv:1908.06849v1eess.SP

TL;DR

Vehicular applications and growing data increase demands for resources, low response time, and bandwidth, while remote-cloud processing causes latency and backhaul pressure. This paper surveys VEC architectures, enabling technologies, research models, applications, and open issues, concluding that the field remains in its infancy with many questions unresolved.

  • Problem

    Growing vehicular applications and data create increasing communication, computation, and storage demands alongside strict response-time and bandwidth requirements.

  • Method

    The paper provides a comprehensive survey covering VEC architecture, advantages, challenges, application scenarios, research topics, classified literature, and future directions.

  • Results

    The survey organizes existing VEC research across computation offloading, caching, data management, flexible network management, security and privacy, and related application areas.

  • Takeaways & Limitations

    VEC is presented as a research field with potential to address vehicular demands for low delay and high bandwidth, but substantial open questions remain.

Abstract

from arXiv · show

As one key enabler of Intelligent Transportation System (ITS), Vehicular Ad Hoc Network (VANET) has received remarkable interest from academia and industry. The emerging vehicular applications and the exponential growing data have naturally led to the increased needs of communication, computation and storage resources, and also to strict performance requirements on response time and network bandwidth. In order to deal with these challenges, Mobile Edge Computing (MEC) is regarded as a promising solution. MEC pushes powerful computational and storage capacities from the remote cloud to the edge of networks in close proximity of vehicular users, which enables low latency and reduced bandwidth consumption. Driven by the benefits of MEC, many efforts have been devoted to integrating vehicular networks into MEC, thereby forming a novel paradigm named as Vehicular Edge Computing (VEC). In this paper, we provide a comprehensive survey of state-of-art research on VEC. First of all, we provide an overview of VEC, including the introduction, architecture, key enablers, advantages, challenges as well as several attractive application scenarios. Then, we describe several typical research topics where VEC is applied. After that, we present a careful literature review on existing research work in VEC by classification. Finally, we identify open research issues and discuss future research directions.

I. INTRODUCTION

Vehicular Edge Computing (VEC) integrates Mobile Edge Computing with vehicular networks to address growing resource demands, latency, and bandwidth pressures. The paper surveys VEC architectures, research topics, existing models, and open directions.

  • Motivation: Vehicular applications require substantial communication, computation, and storage resources, while vehicles face limited capacities and strict delay requirements.These demands arise as vehicular applications and generated data continue to expand.
  • Motivation: Mobile Cloud Computing can help with resource demands, but remote clouds introduce high transmission latency and substantial backhaul bandwidth consumption.Centralizing vehicular data processing at the cloud increases transmission distance and network competition.
  • Survey Scope: The survey reviews VEC architecture, advantages, challenges, application scenarios, and research models covering offloading, caching, data management, network management, and security.The paper also identifies open issues and discusses future research directions.
  • VEC Overview: VEC integrates MEC into vehicular networks by moving communication, computing, and caching resources closer to vehicular users.Its distinguishing environment includes fast vehicle mobility, dynamic topology changes, and rapidly varying channel conditions.
  • Survey Organization: VEC research is organized around an overview, typical research topics, classified literature review, and discussion of open issues and future directions.The paper presents this organization through its content structure and section roadmap.

A. Vehicular Edge Computing:Architecture

The VEC architecture consists of vehicular terminals, RSUs serving as edge servers, and remote cloud services. These layers combine local vehicular capabilities, edge processing, and broader cloud coordination.

  • Architecture: The VEC architecture has three layers: vehicular terminals as the user layer, RSUs as the MEC layer, and cloud servers as the cloud layer.This layered structure places computation and storage resources across vehicles, roadside infrastructure, and remote cloud services.
  • Vehicular Terminals: Vehicles sense, communicate, and compute, using onboard devices to collect information and exchange data through V2V and V2R links.Vehicles may also transfer computation tasks to edge servers or the cloud.
  • RSUs: RSUs receive and process vehicle information, store data, upload information to the cloud, and provide services such as video streaming, traffic control, and navigation.Their roadside placement supports timely interaction with nearby vehicles.
  • Cloud Layer: Cloud services provide greater computation and storage capacity, broader coverage, global views, and centralized management than edge servers.The cloud receives uploaded information from edge servers and mobile nodes to support global-level decisions.
  • Enabling Technologies: Cloud technology, SDN, NFV, and smart vehicles are identified as enabling technologies for VEC.These technologies contribute cloud resources, network-management capabilities, service flexibility, and vehicular intelligence.

1) Cloud Technology [15]:

Cloud technology supplies substantial computation and storage resources but places services far from users. The section frames VEC advantages around moving functionality toward the edge and supporting resource-constrained vehicles.

  • Cloud Technology: Cloud technology provides powerful computation capacity and storage resources, but its remote deployment incurs long latency.VEC moves cloud functionality toward the network edge to sustain diverse application services.
  • Advantages: VEC can reduce execution time because edge servers are closer to vehicular users than the cloud.This is particularly beneficial for delay-sensitive applications such as safety applications.
  • Advantages: VEC can support energy-constrained electric vehicles as smart-vehicle applications increase energy consumption.The passage describes VEC as helping such vehicles provide sufficient support for these applications.

3) Bandwidth:

VEC brings computation, storage, and caching closer to vehicular users, reducing bandwidth pressure and supporting timely services, but mobility and obstacles complicate connectivity.

  • VEC moves computation and storage resources to the network edge, alleviating backhaul bandwidth stress from explosive vehicular data growth.
  • Caching at nearby edge servers lets vehicular users access stored data promptly while reducing storage burden on the remote cloud.
  • Proximity enables edge servers to process vehicle-sensing data into HD maps and deliver them to vehicles.
  • Edge servers can use real-time vehicle, traffic, and network information to deliver content based on user interests.
  • Frequent handovers caused by high vehicle mobility increase delay and can degrade service continuity and user experience.
  • Urban obstacles such as trees and buildings can hinder data transmission, while time-varying channels are difficult to characterize.

3) Resource Management:

VEC resource management must address limited edge capacity, dynamic task and network conditions, security concerns, and diverse application scenarios requiring timely processing.

  • VEC has limited computation and storage resources relative to cloud computing, making resource allocation important under dynamic demands and traffic conditions.
  • Computation-intensive and delay-sensitive tasks may be offloaded to edge servers, but task migration must account for changing channels and topology.
  • VEC must address security and privacy when mutually untrusted vehicles access shared physical edge servers.
  • Edge servers can analyze forwarded vehicle and roadside-sensor data in real time and warn nearby vehicles about risks.
  • Cooperative caching among edge servers and vehicles can let users fetch popular videos directly, decreasing delay and improving experience.
  • Servers can collect vehicle status and environmental information to understand local traffic conditions and control traffic flow to avoid congestion.

4) Path Navigation:

VEC supports navigation, autonomous driving, computation-intensive applications, and data exploitation by supplying nearby computation and storage resources for vehicular services.

  • VEC provides computation and storage resources for real-time navigation, whose implementation requires sensing, collection, and processing.
  • Ultra-low latency and high reliability in VEC can support autonomous driving, including timely environmental understanding and vehicle operations.
  • Computation-intensive applications such as augmented reality and face recognition can be migrated to resource-rich edge servers.
  • Deep exploitation of data generated by sensors and shared among vehicles can provide learned knowledge for data efficiency and network performance.
  • VEC research topics include task offloading, caching, data management, flexible network management, and security and privacy.
  • Cloud distance creates latency and bandwidth pressure for delay-sensitive services and massive application-generated data, motivating VEC.

A. Server-based Offloading

Server-based offloading uses nearby RSUs and other network entities to process vehicular tasks, while mobility, load, incentives, and resource scarcity shape allocation strategies.

  • A. Server-based Offloading: RSUs can act as edge servers that process offloaded tasks without requiring access to the remote cloud.
  • A. Server-based Offloading: Game theory supports joint offloading and channel-selection decisions for multiple vehicles sharing an edge server.
  • B. Vehicle-based Offloading: Unused vehicle resources can assist other vehicles through collaborative offloading aimed at reducing computation and communication delay.
  • C. Mobility Awareness: Mobility-aware migration can transfer tasks to a new edge server or forward completed results when vehicles leave the original server’s coverage.
  • C. Mobility Awareness: Predictive-model transmission and prediction-based offloading address task requirements and vehicle mobility to improve transmission efficiency and satisfy delay requirements.
  • C. Mobility Awareness: Multi-server resource allocation can jointly balance load, offloading choices, vehicle requirements, and mobility.
  • D. Incentive Strategy: Incentive schemes compensate resource providers because self-interested entities may otherwise withhold unused resources.
  • D. Incentive Strategy: Contract-theoretical offloading can maximize edge-server revenue while improving vehicle utilities.

E. Energy Efficiency

Vehicular edge caching addresses growing data traffic and latency demands by placing content in edge servers or vehicles, while caching decisions must account for constrained resources and mobility.

  • Caching motivation: Edge caching uses storage at edge servers and vehicles to reduce data traffic and latency for vehicular applications.The edge network supplements the cloud with storage resources close to users.
  • Caching design: Caching decisions concern where to cache, which contents to select, and which policy objective to optimize.Objectives include offloading traffic, reducing delay, improving QoE, and minimizing energy consumption.
  • Caching at RSUs: Limited RSU storage and communication range constrain caching decisions and require effective selection of cached contents.Competition among content providers can further complicate storage utilization.
  • Mobility challenges: Vehicle mobility causes unstable connectivity, making continuous content transmission difficult when requests span multiple RSU coverage areas.A vehicle may pass through several RSUs before obtaining the requested content.

B. Caching at Vehicles

Caching at vehicles exploits underused vehicle storage and cooperation to improve content availability, but must adapt to limited capacity, changing popularity, and mobility.

  • Cooperative caching: Vehicles can act as caching nodes, and cooperative caching can improve hit ratio and reduce delay by sharing unused storage.Parked vehicles are described as useful storage providers because of their large number and long dwell time.
  • Cooperative caching: Vehicle-based cooperative caching supports data dissemination when fixed RSUs are sparse or poorly matched to highly mobile vehicles.One cited approach shares traffic information among vehicles through P2P communication.
  • Caching challenges: Vehicle caching must address limited storage capacity, popularity calculation, and vehicle movement.Proposed schemes consider cluster-wide caching status, content and chunk reference frequency, and mobility-aware data forwarding.
  • Layered caching: Layered caching across BSs, RSUs, and vehicles can increase content availability for users on the move.Placement across vehicular and RSU layers has been studied with delay and QoE objectives.
  • Popularity prediction: Vehicle mobility and changing user interests make content popularity context-dependent and unknown before caching.Relevant contexts include location, network topology, and personal status.
  • Proactive caching: Proactive caching combines vehicle mobility prediction with learning to balance user QoS against RSU caching cost.A cited scheme uses an LSTM for mobility prediction and Q-learning with a deep learning network.

VI. DATA MANAGEMENT IN VEHICULAR EDGE COMPUTING

Vehicular edge computing addresses data-management demands created by abundant sensing and content-sharing data, emphasizing collection, processing, dissemination, and resource-aware delivery.

  • Motivation: Vehicular applications generate large data volumes that create response-time and backhaul-bandwidth challenges.Vehicle devices produce data, while wireless communication also enables data sharing among vehicles.
  • Motivation: Effective data management must collect, analyze, process, and disseminate data generated and consumed by vehicles.Deeper exploitation of these data is presented as a way to obtain knowledge for data efficiency and network performance.
  • Data collection: Vehicular sensing frameworks use adaptive pull and push collection, while vehicular micro clouds process and aggregate data before delivery to a data center.These approaches combine fog computing with vehicular sensing or virtual edge servers.
  • Data dissemination: MEC-assisted publish/subscribe dissemination targets stringent delay and reliability requirements of delay-sensitive applications such as safety messaging.Safety messages require rapid and precise forwarding to enhance road safety and avoid potential danger.
  • Content delivery: Content delivery can reduce remote-cloud access by serving requested contents from nodes that already store them.Related work considers balancing computation, caching, and communication resources and preallocating realtime streaming content.
  • Incentives: Data-sharing schemes must account for selfish nodes by jointly considering vehicle selection, capacity, and incentives for collaborative caching and dissemination.Reported incentive approaches aim to make sharing robust and distributed.

VII. FLEXIBLE NETWORK MANAGEMENT IN VEHICULAR EDGE COMPUTING

Flexible vehicular network management combines SDN, fog or edge computing, and heterogeneous architectures to address mobility, resource allocation, delay, reliability, and security concerns.

  • SDN-enabled management: SDN separates control and data planes, enabling flexible network management, lower management costs, and global network visibility.The data plane performs operations commanded by a controller in the control plane.
  • Mobility management: SDN and fog-computing vehicular architectures address mobility through handover management, hybrid access, and delay-oriented resource allocation.These designs target low delay and high reliability.
  • Heterogeneous architectures: MEC-assisted SDN architectures target required data ratios, reliability, scalability, responsiveness, and improved network management in heterogeneous vehicular networks.One architecture combines SDN and MEC for 5G-enabled software-defined vehicular networks.
  • Resource management: Vehicular architectures jointly optimize networking, storage, and computation resources to alleviate traffic congestion and improve resource management.Programmable SDN control is used to support network optimization.
  • Resource management: Task offloading and fog-server selection can be coordinated with resource allocation to increase the probability of completing tasks within a given period and meet delay requirements.SDN assistance is used in both task offloading/resource allocation and fog-server assignment.
  • Security and privacy: Security and privacy remain deployment concerns because mobility and device heterogeneity complicate trust and authentication, while malicious or faulty nodes threaten data integrity and service availability.Obstacles such as buildings, trees, and trucks can obstruct communication line of sight and affect reliability.

2) Confidentiality and Integrity:

VEC security and privacy research addresses threats to RSUs, offloaded data, navigation, sensing, location sharing, and blockchain-based trust and data management.

  • RSU security: RSUs deployed in public places can be eavesdropped on or attacked after adversaries infer their locations from traffic statistics.Dummy traffic generated by vehicles is proposed to mislead traffic statistics and protect RSUs from service-disruption attacks.
  • Data confidentiality: Sensitive data offloaded to edge servers creates confidentiality risks, including exposure of location, purchase history, and healthcare records.Physical-layer security is investigated for secrecy provisioning when tasks are offloaded over RF channels.
  • Privacy-preserving computation: Resource-constrained vehicles may be unable to perform computation-intensive cryptographic operations locally.One approach migrates encryption and decryption to fog and cloud servers while preserving confidentiality, privacy, anonymity, and unlinkability.
  • Privacy-sensitive applications: Secure navigation and mobile crowd sensing require privacy protections because traffic, location, and sensory data can expose users or create safety risks.The surveyed work includes secure privacy-preserving navigation and fog-based privacy protection for mobile crowd sensing.
  • Location privacy and trust: Vehicular communication and location-based services make vehicles trackable through communication and movement behaviors.Blockchain-based methods are explored for decentralized trust management, secure data sharing, reputation assessment, and privacy-preserving carpooling.

IX. OPEN RESEARCH ISSUES AND FUTURE WORK

Open VEC research must address heterogeneous application requirements, dynamic vehicle distributions, resource-sharing incentives, and security constraints. These challenges arise alongside fast mobility and harsh channel conditions, while the field remains early-stage and requires further investigation.

  • Flexible resource scheduling: Flexible scheduling must distinguish safety applications with strict delay requirements from non-safety applications that tolerate some delay.Examples include collision avoidance and traffic control for safety, versus multimedia downloading for non-safety use.
  • Dynamic network conditions: VEC resource and connectivity management must accommodate heterogeneous tasks, varying vehicle densities, and uneven spatial distributions of vehicular users.Applications require low delay, high reliability, computation capacity, and storage space, while density changes over time and across areas.
  • Incentive mechanisms: Effective resource sharing requires pricing mechanisms that value resources and balance profits between consumers, providers, and related entities.The framework also raises questions about dividing retained profits among participating entities.
  • Security and privacy: Offloaded tasks require confidentiality and integrity protections because they may contain sensitive data and can be modified during forwarding or storage.The section identifies verifiable computing as relevant to these requirements.
  • VEC-specific challenges: VEC extends cloud functions to the network edge, but fast vehicle mobility and harsh channel environments create challenges distinct from traditional MEC.These intrinsic vehicular-network features contribute to dynamic operating conditions for edge services.
  • Research agenda: The paper surveys VEC architecture, enablers, applications, research topics, classified literature, open issues, and future directions.The classified review covers task offloading, caching, data sharing, flexible network management, and security and privacy.
Loading 1908.06849v1…