Vehicle-to-Everything Services in 3GPP 5G Networks: An Empirical Analysis

5/6G vehicle-to-everything (V2X), particularly vehicle-to-network (V2N) communication, should provide real-time data transfer between vehicles and the cloud. In this paper, we evaluate the capability of direct commercial 5G V2N connectivity to support 5G V2X services. We perform real-world local and cross-country drive tests to measure key performance metrics such as throughput and endto-end latency. We use the measurements to determine whether the 5G V2N meets the 3GPP standard requirements for V2X services. We then assess the feasibility of V2X services based on compliance with related latency, and throughput thresholds. Finally, we derive the signal strength thresholds required to support V2X services with low-level and high-level automation, providing insights for future 6G development.

Introduction

As intelligent vehicle users increases and demanding applications increase, the requests for network connectivity and other cloudnative services increase. These services include but are not limited to navigation, route optimization, over-the-air updates, real-time alerts, infotainment, and leisure functionalities. The future softwaredefined vehicles have been foreseen to enable various services, applications, and innovations on automotive boards [6]. Most require large-scale data, access to effective artificial intelligence models, and back-end services. To enable reliable network access to the cloud, considerations have turned into vehicle-to-everything (V2X) service-oriented systems [20]. However, the capability of the current commercial 5G networks does not seem to meet the latency requirements [27]. In this paper, we address the service availability of commercial 5G for V2X service-oriented systems using empirical data and experimental methods.

Unlike regular smartphone users in more static conditions, vehicles require fast and stable network connections to ensure safety and security in transportation. Prioritizing the vehicular network traffic ensures that critical safety messages, collision, and weather warnings are transmitted immediately. In this context, leveraging cellular vehicle-to-everything (C-V2X) [7] communication technologies is required, with vehicle-to-network (V2N) integration playing a pivotal role in enabling reliable, real-time data exchange between vehicles and the cloud. As a result, Society of Automotive Engineers (SAE) Level 3 [10] and above benefits greatly from V2X with minimal downtime, latency, and continuous availability in vehicle autonomy and functionality. V2N is a type of C-V2X connectivity that allows vehicles to connect to the Internet and access cloud services through cellular networks using the Uu interface[7], which operates in the traditional mobile broadband spectrum. While time-critical actions, like collision avoidance, need to be handled onboard, non-time critical services, such as facilitating warning alerts and risk avoidance that support reliable information sharing require reliable V2N.

The primary objective of our work is to determine the capability of direct 5G V2N to provide standard V2X services for vehicular users. Our main contributions are as follows:

1. V2N connectivity over dedicated vehicle-to-vehicle or vehicle-to-infrastructure: Related works [14, 15] have mostly focused on engaging local networks with vehicle-to-vehicle (V2V) or vehicle-to-infrastructure (V2I). We specifically examine the potential of commercial 5G V2N as a direct communication link to enable both V2N and V2V/V2I-assisted services. V2N uses existing cellular infrastructure to connect vehicles to the cloud, a more scalable solution than deploying dedicated V2V/V2I infrastructure.

2. Real-world 5G V2N capability against 3rd Generation Partnership Project (3GPP) [1] standards: We use real-world drive tests in three countries to measure the current 5G V2N performance. This provides a practical perspective on the capabilities and limitations of the existing 5G V2N services and highlights needed future developments.

3. Signal strength thresholds for automation zones: We propose reference signal received power (RSRP) coverage thresholds 𝛼 for low-level automation and 𝛽 for high-level automation within the coverage limit 𝛾 (Figure 1). These zones can inform vehicles about enabled V2X, such as platooning and safety alerts in low automation zones. High automation zones can support advanced, latency-sensitive services like remote driving and AI-aided decisionmaking via the vehicle-to-cloud continuum.

4. How vehicle mobility impacts signal strength: We assess how vehicle speed, environmental type, network deployment mode, and handovers influence signal strength. Our results indicate a need to improve network reliability, expanding the availability of 5G stand-alone (SA) and upcoming 6G access points in road areas. This will enhance connectivity for dynamic users amid varying vehicle speeds and handovers.

Related Work

The 3GPP [1] has proposed network service requirements for V2X use cases such as vehicle platooning, advanced driving, extended sensors, and remote driving [3] (see Table 2). These service requirements for V2X services are defined based on automation levels, ranging from "lowest" to "highest", and represent the degree to which a vehicle can perform tasks like collision avoidance and cooperative driving based on connectivity. In this study, we categorized the automation levels into low and high by grouping the "lowest", "lower", and "low" levels of automation as "low" and the "high", "higher", and "highest" as "high". This is used to streamline the analysis and compare the test outcomes with the defined requirements. For each category, the minimum limits for E2E latency and throughput are specified, applying to all link types: downlink, uplink, and sidelink.

By meeting the standard service requirements, V2N replacing V2I/V2V is a possibility due to technologies like software-defined networking, network slicing, and network virtualization, which can prioritize vehicle traffic, throughput within the network, and reducing latency [20]. Such a shift could reduce costs associated with deploying and maintaining separate 5.9 GHz-based V2I infrastructure. Hence, the V2N’s ability to cater to V2I or V2I mode-based services must be evaluated through experimental studies, and the V2N limitations must be identified.

Some previous research [25, 26] proposes analytical models for 5G V2X services using simulated environments. There is a lack of experimental evaluations of commercial 5G V2N for V2X, although previous works [12, 23] have experimentally studied the 4/5G V2V/V2I performance in terms of throughput, latency. Similarly, previous works [5, 8] mentioned the usefulness in determining signal strength thresholds for V2I/V2V assisted V2X services, but they have not been defined for V2N. Works on the empirical performances in 5G [13] have focused on network performance in urban and highway scenarios and cross-border environments in [9]. However, there works on V2N capabilities to provision V2X services is lacking, and it is an open research question if there is a possibility of replacing V2I or V2V while maintaining comparable performance. The existing literature has not defined signal strength thresholds for 5G V2N, based on real-world measurements, to support V2X services and automation levels (DoA). Table 1 compares major experimental works on 5G V2X service capabilities and challenges, highlighting the focus in this study.

Experimental Setup and Measurements

tldr

Performance Analysis

tldr

Discussion

We studied the current 5G V2N capability, minimum required signal strength, and limitations to provide V2X network services according to the 3GPP standard requirements. This is needed to determine the network constraints when designing cellular network-assisted vehicular system architectures. As for the existing capabilities, the results showed that meeting the strict latency requirements, <20 ms, and throughput requirements, ≥ 1000 Mbps, for V2X services via 5G V2N remains challenging. Specifically, remote driving, proposed in 3GPP to operate primarily via V2N, is less feasible via the current 5G. However, with 6G, the quality of service for V2X is expected to improve significantly, providing standard performance in latency <1 ms and throughput ≥ 100 Gbps [19]. We defined the RSRP signal strength thresholds required to support low and high-level DoA services based on real-world data. These thresholds can support designing hybrid V2N and V2V connectivity systems.

Limitations: This study conducted cellular network-specific characteristics such as latency, throughput, and RSRP using smartphones instead of in-car antenna. Smartphone-assisted in-car measurements could represent the possible worst-case connectivity conditions, as a smartphone’s performance is limited by lowerquality antennas, shielding from the car’s body, and orientation. In contrast, dedicated V2N hardware typically integrates receiving antennas optimized for dynamic vehicular environments with varying signal strengths and provides better signal reception. Primarily, HTTP-only UL/DL tests were conducted to assess throughput required for accessing edge/cloud servers in V2X-assisted driving. Cross-country tests started from Oulu, Finland (City & Country A), via Kiruna in Sweden (City & Country B), ended in Lødingen, Norway (City & Country C), which may affect global result generalization. 99% of the time, the device connected to the network via 5G NSA. The results represent the NSA performance and may have resulted in higher delays and lesser throughput than the 5G SA connectivity.

Future impact: As 5G SA networks aim to deliver better throughput and E2E latency than NSA, 5G V2N capability evaluation under roadside 5G SA sites is a future focus. Improving handover efficiency and designing 5G V2X architectures that support switching between V2X modes based on signal strength is a future direction. Edge caching, and computing assisted architectures may reduce the latency. Our driving test datasets, which are related to developing V2N-assisted analytical models or systems, is shared on GitHub, eliminating the need for retesting.

Conclusion

We assessed the capability of direct 5G V2N to meet 3GPP network requirements for V2X services through local, and cross-country drive tests. The 5G V2N RSRP for minimum connectivity, low-level DoA, and high-level DoA were -119.5, -112.29, and -102.86 dBm, respectively. We observed that 5G V2N could provide low-level DoA V2X services up to 75% time, potentially serving as an alternative to V2I. Hence, the current 5G V2N can support services that require latency 30 < and < 100 ms and throughput < 500 Mbps. However, 5G V2N struggles to support V2X services that require latency < 20 ms and throughput > 1000 Mbps.