Written by Ramon Sanchez-Iborra, Jorge Gallego-Madrid, and Antonio Skarmeta
Recent advances in networking and computation technologies are facilitating the resurge of a hyper-connected vehicular ecosystem. First, the flexibilities provided by related technologies such as Software Defined Networking (SDN) and Network Function Virtualization (NFV) permit the rapid reconfiguration of communication infrastructures by means of virtualization techniques. Besides, the availability of multiple Radio Access Technologies (RATs) expands the range of vehicular applications that can be supported in a plethora of scenarios. However, the arrival of new applications calls for the development of novel architectures capable of orchestrating both communication and computation resources to cope with their stringent requirements. This article presents a novel holistic architecture that provides support to next-generation vehicular applications. It encompasses multiple communication technologies such as 5G, as well as others related to the vehicular and Internet of Things (IoT) paradigms. Besides, it also presents multiple distributed processing sites which enables a flexible fog-edge-core-cloud configuration making such implementation particularly suited for supporting vehicular communications. This testbed aims to be a reference in its class as a living lab, open to support state-of-the-art experiments and innovations.
Vehicular services and applications have received great attention in recent years given their great potentials in improving the experience of vehicle’s occupants ranging from assisted driving to infotainment services. However, there have been technological limitations that have slowed down the deployment of these attractive developments until recent years with the deployment of 5G cellular networks worldwide. On the one hand, from a data communication perspective, the first approach for providing vehicles with connectivity was the implementation of IEEE 802.11p-based solutions. The limited transmission range of earlier technologies and the consequent need of deploying a complex and expensive fixed infrastructure notably hindered the implementation of functional infrastructures covering large areas. Another technical feasibility consideration is related to the computational processing needs, the dynamic and distributed nature of vehicular application scenarios poses great challenges for the adequate placement of processing functions within the fixed infrastructure that could meet the strict Quality of Service (QoS) requirements of vehicular applications.
The arrival of technologies such as 5G, IoT, SDN, NFV, used in conjunction with Artificial Intelligence (AI) and Machine Learning (ML) solutions, among others, is shaping smarter connected vehicles. However, there is still a lack of management platforms that could orchestrate and coordinate these technologies aiming at supporting truly implementable vehicular services and applications with interoperability issues yet to be resolved. A wide variety of applications have been made possible in recent years , emerging services such as higher level of autonomous driving, enhanced navigation, comprehensive infotainment services, optimized traffic management, as well as improved safety features. Until the most futuristic ones such as in-vehicle holographic services, augmented/mixed/virtual reality (XR), brain-vehicle interface, etc. are all important elements in making connected vehicles smarter.
In this article, we present a novel holistic platform encompassing both communication and computation resources that focuses on supporting next-to-come applications in heterogeneous vehicular scenarios. This rich infrastructure is being deployed at the Espinardo Campus of the University of Murcia (Spain). As shown in Figure 1, it presents a distributed processing infrastructure and different RATs to support several vehicular test cases.
Figure 1. Holistic communication and computation infrastructure at the University of Murcia (Spain).
CommunicationsConsidering different transmission alternatives is crucial to provide connectivity continuity along the covered area. 5G is taken as the base technology but it is enriched with other solutions that provide diverse features. Thus, the broadband and low-latency communications enabled by three 5G Stand Alone (SA) access points are complemented with the long range and reliability of LoRa, and the Vehicle-to-Infrastructure (V2I) and Vehicle-to-Vehicle (V2V) capabilities enabled by 802.11p-based communications. This heterogenous set of RATs permits the implementation of applications and services with different QoS characteristics. Furthermore, the selection of RAT to be employed by a vehicle in a certain moment, attending to contextual reasoning, is a hot topic of study . Finally, considering the backbone infrastructure, different programmable elements such P4 switches or enhanced Berkeley Packet Filter (eBPF)-enabled nodes are deployed to provide re-programmability to the data-plane which is managed using state-of-the-art SDN controllers. This flexibility and control over the user traffic permit the configuration of end-to-end network slices through dynamic resource allocation.
As mentioned previously, correct placement and orchestration of network functions is crucial to reach the high QoS requirements of vehicular applications. For that reason, the infrastructure presents two different sites to provide edge and core processing provisioning. Besides, to manage and orchestrate the dynamic deployment of VNFs, processing functions and micro-services within the computation infrastructure, a Network Function Virtualization Management and Orchestration (NFV MANO) instance is operative. The NFV MANO automatically manages the physical infrastructure through a series of Virtual Infrastructure Managers (VIMs), making it possible to onboard, deploy and evaluate the instantiated applications. It also enables the autonomous and dynamic reconfiguration of the network resources, by deploying end-to-end network slices to handle the characteristic changing environment of vehicular scenarios. These network resources are accessible from the radio network infrastructure and provide different QoS performance in real-time depending on the employed access and the network conditions. Finally, a top-level orchestration layer is in charge of monitoring and coordinating all the resources described above by means of AI/ML-powered mechanisms. Figure 2 presents the logical configuration of the components constituting the architecture.
Figure 2. Logical components constituting the vehicular communications architecture.
As the most innovative communication and computation technologies continue evolving, holistic platforms gathering these solutions are needed to serve as valid testbenches for critical applications such as the vehicular ones. This article presented a rich infrastructure that counts with several RATs, computing, and network resources that are dynamically managed and orchestrated, which permits the deployment and evaluation of a large set of current and future vehicular applications. Besides its exploitation in numerous international projects, the infrastructure is open to receive researchers and practitioners interested in testing novel developments within the vehicular ecosystem. Such platform is expected to open up numerous opportunities in integrating connected vehicles into the broader smart city environment.
- H. Guo, X. Zhou, J. Liu, and Y. Zhang, “Vehicular intelligence in 6G: Networking, communications, and computing,” Vehicular Communications, vol. 33, p. 100399, Jan. 2022, doi: 10.1016/j.vehcom.2021.100399.
- R. Sanchez-Iborra, L. Bernal-Escobedo, and J. Santa, “Machine learning-based radio access technology selection in the Internet of moving Things,” China Communications, vol. 18, no. 7, pp. 13–24, Jul. 2021, doi: 10.23919/JCC.2021.07.002. IEEE Wireless Communications, vol. 24, no. 6, pp. 14-21, Dec. 2017, doi: 10.1109/MWC.2017.1600414.
This article was edited by Bernard Fong