PaperlayerView exact revision on arXiv

Source-linked AI summary

Software Defined Space-Air-Ground Integrated Vehicular Networks: Challenges and Solutions

Ning Zhang, Shan Zhang, Peng Yang, Omar Alhussein, Weihua Zhuang, Xuemin Shen

arXiv:1703.02664v2cs.NI

TL;DR

Terrestrial vehicular networks face coverage, capacity, and mobility constraints, while diverse services call for complementary network capabilities. The paper proposes an SDN-based space-air-ground architecture with segment controllers and higher-tier orchestration; its case study reports stronger signal strength for coordinated links than greedy and random establishment, in a one-plane study.

  • Problem

    Terrestrial vehicular networks face costly deployment, poor rural coverage, urban capacity constraints, and challenges supporting highly mobile vehicles.

    Source: §I · p. 2

  • Method

    The proposed SSAGV architecture uses segment-specific SDN controllers and higher-tier controllers to orchestrate space, air, and ground network operation.

    Source: A. Network Architecture · p. 7

  • Results

    The SDN-enabled coordination scheme outperforms individual greedy and random link establishment in average signal strength across the reported beam-count and HAP-count comparisons.

    Source: B. Simulation Results · p. 14

  • Takeaways & Limitations

    The architecture is presented as a way to exploit space, air, and ground advantages for diverse vehicular services and support agility, flexibility, and resource sharing.

    Source: §VII · p. 16

    The case study demonstrates performance gain for one satellite plane; the passage says gains in a practical network could be more appealing.

    Source: B. Simulation Results · p. 14

Abstract

from arXiv · show

This article proposes a software defined space-air-ground integrated network architecture for supporting diverse vehicular services in a seamless, efficient, and cost-effective manner. Firstly, the motivations and challenges for integration of space-air-ground networks are reviewed. Secondly, a software defined network architecture with a layered structure is presented. To protect the legacy services in satellite, aerial, and territorial segments, resources in each segment are sliced through network slicing to achieve service isolation. Then, available resources are put into a common and dynamic space-air-ground resource pool, which is managed by hierarchical controllers to accommodate vehicular services. Finally, a case study is carried out, followed by discussion on some open research topics.

I. Introduction

Connected vehicles support safety, infotainment, and location-dependent services, but terrestrial networks face coverage, capacity, deployment, and mobility challenges. Integrating satellite, aerial, and terrestrial networks can exploit their different strengths for vehicular services.

I. Introduction

  • Connected vehicles enable safety, infotainment, and location-dependent services through communication with neighboring cars, roadside infrastructure, and traffic control centers –.Examples include collision avoidance, online gaming, and route optimization.Source: §I · p. 1
  • Terrestrial 802.11p infrastructure is costly and slow to deploy, while LTE has rural coverage and urban capacity constraints; both also face handover challenges for highly mobile vehicles.The passage identifies frequent handovers as vehicles associate with new roadside units or base stations.Source: §I · p. 2

A. Background

  • Terrestrial networks can provide high data rates, satellites broad coverage, and HAPs regional capacity with incremental deployment and on-demand repositioning.Satellite coverage is suited to rural and remote areas, while HAPs can supplement hotspot capacity.Source: A. Background · pp. 3–4

B. Motivations for Integration

Standalone terrestrial networks struggle to meet diverse vehicular service requirements cost-effectively and flexibly. Interworking can combine the segments' complementary coverage, capacity, and broadcast capabilities.

B. Motivations for Integration

  • Rural areas and highways have poor terrestrial coverage, and extending infrastructure in sparsely populated areas is costly.Satellite systems or HAPs can provide connectivity over large areas; a satellite cell can span several hundred kilometers.Source: B. Motivations for Integration · p. 4
  • Frequent handovers between small terrestrial cells degrade on-the-move service performance for vehicles on highways.The passage gives LTE macrocell coverage of around 1 km and small-cell coverage below 300 m [10].Source: B. Motivations for Integration · p. 4
  • A single technology cannot efficiently serve diverse vehicular services: LTE supports high-rate unicast, while satellite and HAP systems are more efficient for broadcasting similar content to many users.Source: B. Motivations for Integration · p. 5
  • Traffic-load dynamics can cause urban or suburban congestion, while adding terrestrial resources for peak demand is not cost-effective; HAP resources can accommodate bursty demand.Source: B. Motivations for Integration · p. 5
  • Interworking can pair terrestrial high-rate service in urban and suburban areas with satellite rural coverage and HAP hotspot capacity or emergency communications.Satellites and HAPs can also relay road and geolocation information and offload terrestrial demand.Source: B. Motivations for Integration · p. 5

C. Challenges

Space-air-ground integration must accommodate heterogeneous, changing resources while preserving legacy services and supporting vehicular QoS. Interoperability, management complexity, and mobility-driven dynamics also complicate operation.

C. Challenges

  • Interoperation is limited by differing standards, links, and devices, as well as closed systems and dedicated gateways needed for protocol and format conversion.Source: C. Challenges · p. 5
  • Managing many devices with dedicated control interfaces and varied hardware and software makes flexible reconfiguration and responses to congestion or link failures difficult.Source: C. Challenges · pp. 5–6
  • Satellite and HAP mobility changes resource availability, while vehicle mobility creates time-varying, geographically uneven demand.LEO satellites orbit in periods under 130 minutes.Source: C. Challenges · p. 6
  • Integrated networks must support diverse QoS requirements despite heterogeneous resources and availability that changes as segments prioritize their legacy services.Legacy services are existing segment services, such as satellite earth observation and navigation, rather than vehicular services.Source: C. Challenges · p. 6

III. Software Defined Space-Air-Ground Integrated Vehicular Networks

The proposed SSAGV architecture applies SDN to address integration challenges by separating control and data planes and enabling logically centralized, programmable network control.

III. Software Defined Space-Air-Ground Integrated Vehicular Networks

  • SDN separates the control plane from the data plane and provides logically centralized control with a global network view.Source: §III · p. 7
  • Open interfaces support network programmability and reconfiguration, enabling a dynamic, manageable, cost-effective, and adaptable network.Source: §III · p. 7
  • The paper proposes an SDN-based space-air-ground integrated vehicular architecture to address the identified challenges.Source: §III · p. 7

A. Network Architecture

The SSAGV architecture coordinates space, air, and ground segments through segment-specific SDN controllers and higher-tier controllers for vehicular services. Network slicing isolates legacy services while making remaining resources available to vehicular services through a shared, dynamic pool.

A. Network Architecture

  • Higher-tier SDN controllers coordinate the space, air, and ground segments, whose dedicated interfaces address their distinct standards and devices.Different tiers use abstracted network information, including vehicle density, content popularity, and active-user counts.Source: A. Network Architecture · p. 7
  • Network slicing partitions each segment’s resources into isolated service slices, with hypervisors coordinating local scheduling and network-wide allocation.Resources are allocated to legacy services, including earth observation, navigation, and weather monitoring.Source: A. Network Architecture · pp. 7–8
  • Remaining resources from each segment enter a shared space-air-ground pool for vehicular services, and slicing changes dynamically with service needs.Source: A. Network Architecture · p. 8

B. Layered Structure

The architecture separates infrastructure, control, and application functions. Virtualized shared resources and hierarchical controllers support vehicular services and network management across heterogeneous segments.

B. Layered Structure

  • The infrastructure layer combines computing, storage, and communication resources into a dynamic pool, while virtualization abstracts logical resources from physical ones.Services can run on collections of virtualized resources rather than fixed physical resources.Source: B. Layered Structure · p. 8
  • Hierarchical SDN controllers coordinate local through global domains, using segment-specific southbound interfaces and upper-tier coordination.Interfaces support functions such as satellite beam steering, UAV movement control, and LTE resource-block allocation.Source: B. Layered Structure · p. 9
  • The application layer uses northbound interfaces to translate service requests into controller rules and device instructions.Its functions include mobility management, traffic classification and engineering, and network hypervisors for virtual networks.Source: B. Layered Structure · p. 9
  • The proposed architecture is presented as offering simpler management and upgrades, optimized resource use, and adaptable control as network conditions change.The stated dynamics include topology changes and link failures.Source: B. Layered Structure · pp. 9–10

IV. Network Operation

The operation model connects infrastructure providers, vehicular service providers, and customers, then uses hierarchical controllers to manage a large, dynamic network at different spatial and temporal scales.

A. Working Relation

  • Infrastructure providers supply resources to service providers, which provision services to customers; brokers can negotiate across providers when resources are insufficient.Providers may also form coalitions to meet service requirements, such as HAP owners cooperating to supply capacity in a target area.Source: A. Working Relation · p. 10

B. Hierarchical Network Operation

  • Because the network spans large scale, diverse services, heterogeneous resources, and dynamic states, operation is organized hierarchically across spatial domains.Upper-tier controllers coordinate lower-tier controllers, which manage devices in smaller areas.Source: B. Hierarchical Network Operation · p. 10
  • Upper-tier controllers make coarse-grained decisions such as steering traffic or changing base-station modes, while lower-tier controllers handle fine-grained device control and immediate requests.The tiers adapt at different time scales: upper tiers use high-level network status, while lower tiers respond to instantaneous user requests.Source: B. Hierarchical Network Operation · p. 11
  • Controllers use information abstracted for their tier, such as user density for upper tiers and detailed user locations for lower tiers.Source: B. Hierarchical Network Operation · p. 12

C. Big Data-Assisted Operation

Big data-assisted operation collects vehicle-related information across ground, air, and space segments, processes it, and analyzes network states to inform controller decisions.

C. Big Data-Assisted Operation

  • Ground data can come from roadside cameras, electromagnetic transducers, and acoustic sensors, while aerial photography and videos provide road information from air or space.Source: C. Big Data-Assisted Operation · p. 12
  • Collected raw data is processed before transmission through methods such as compression and fusion, and satellites with onboard processing can extract useful information.Source: C. Big Data-Assisted Operation · p. 12
  • Data mining and machine learning analyze network states, while predictions of traffic, content popularity, and user mobility can inform decisions.The passage describes big data and SDN as enabling intelligent and automated network operation.Source: C. Big Data-Assisted Operation · p. 12
  • Fig. 4 is identified as a performance comparison in terms of average signal strength.Source: Figure 4 · p. 13

V. A Case Study

The case study evaluates SDN coordination of satellite-to-HAP links for vehicular content delivery when direct satellite-to-ground optical links are poor. With real-time spatial information, the scheme outperforms greedy and random link establishment in the simulated settings.

A. Network Setting

  • The simulated setting uses a 48-satellite Globalstar LEO network in 8 planes and HAPs uniformly distributed at 20 km altitude.The LEO satellites orbit in under 130 minutes, while HAPs remain relatively stable relative to Earth; the minimum communication elevation angle is 10°.Source: A. Network Setting · p. 13
  • The case study uses HAPs as relays when weather makes direct satellite-to-ground optical links poor, connecting satellites by optical links and ground users by microwave links.Source: A. Network Setting · p. 13

B. Simulation Results

  • The coordination problem is one-to-many matching between satellites and HAPs, using real-time satellite and HAP positions available to the SDN controller.Legacy systems without spatial information make myopic greedy decisions based on high elevation angles; random link establishment is another benchmark.Source: B. Simulation Results · pp. 13–14
  • The SDN-enabled scheme outperforms greedy and random link establishment with 6 satellites and 50 HAPs, quickly converging to the optimal value.The comparison uses average signal strength versus the maximum number of satellite beams.Source: B. Simulation Results · p. 14
  • With each satellite limited to 3 links, SDN-enabled coordination still outperforms greedy and random schemes as the number of HAPs increases.Average signal strength decreases as more HAPs compete for the limited satellite links.Source: B. Simulation Results · p. 14
  • The study demonstrates considerable performance gain for one satellite plane; the authors suggest gains in a practical multi-plane network could be more appealing.Source: B. Simulation Results · p. 14

VI. Research Issues

The research issues include implementing SDN control across distinct network segments and allocating heterogeneous, dynamic resources to vehicular services. The authors identify a time-evolving resource graph as one way to represent changing satellite and aerial resources for allocation.

A. Software Defined Platform Implementation

  • SDN implementation requires choosing controller scope and control granularity to balance performance and complexity, while accounting for segment-specific interfaces and features.The deployment of hierarchical SDN controllers also affects network performance and needs further investigation.Source: A. Software Defined Platform… · pp. 14–15

B. QoS-aware Resource Allocation

  • A time-evolving resource graph (TERG) can represent satellite and aerial resource evolution on a unified time-space basis for dynamic vehicular-service allocation.Vertices represent network elements such as satellites and HAPs; edges denote resource availability, and the graph can be combined with available ground resources.Source: B. QoS-aware Resource Allocation · p. 15
  • Efficient resource allocation is challenging because the integrated network is highly dynamic and resources and services are heterogeneous.Allocation should account for satellite trajectory predictions and controllable HAP movement.Source: B. QoS-aware Resource Allocation · p. 15

C. Interaction among Different Parties

The paper identifies coordination among infrastructure providers and service providers, network virtualization, and security as research issues for integrated vehicular networks. Its conclusion calls for further work on these directions while presenting the architecture as a means to support diverse services efficiently and cost-effectively.

C. Interaction among Different Parties

  • Infrastructure providers seek revenue without degrading legacy services, while vehicular service providers seek cost-effective resource use; game theory can model their interaction.The authors identify non-cooperative games for price negotiation and resource dispatch, and auction theory for multiple providers and service providers.Source: C. Interaction among Different Parties · p. 15

D. Network Virtualization

  • Service-oriented virtual networks can be customized to requirements, while dynamic embedding of virtual nodes can stabilize topology despite physical-network changes.Optimal virtual topology, dynamic embedding, and customized end-to-end protocols remain research needs.Source: D. Network Virtualization · pp. 15–16

E. Network Security

  • SDN controllers face DoS and insider-attack risks, while vehicular services require differing confidentiality, integrity, and authentication protections.The authors identify investigation of SSAGV network security as necessary.Source: E. Network Security · p. 16

VII. Conclusion

  • The proposed architecture aims to support diverse vehicular services efficiently and cost-effectively, with agility, flexible management, and resource sharing across segments.The authors call for extensive research on the outlined directions to accelerate SSAGV network development.Source: §VII · p. 16
Loading 1703.02664v2…