Study on frequency requirements for 5G NR-V2X direct communication
[Copy link]
Li Yan, Gao Lu
(Qualcomm Wireless Technologies (China) Co., Ltd., Beijing 100013)
Abstract: This paper studies the spectrum requirements of direct communication in 5G New Radio-Vehicle-to-Everything (NR-V2X) technology for autonomous driving scenarios, and provides the spectrum requirement research method, assumed parameters and evaluation results. In the NR-V2X system, the broadcast mode is used to send messages that carry status information and environmental information. These messages require a frequency of at least 30~40 MHz. The multicast mode of NRV2X can support negotiation and decision-making between autonomous driving groups. Although the multicast mode is more critical for supporting advanced applications, since group communication is basically event-triggered, the total traffic volume transmitted through the multicast mode is far less than the total traffic volume of broadcast messages. In the early stages of NR-V2X frequency research, the frequency requirements of the multicast mode can be temporarily ignored.
Keywords: 5G NR-V2X; broadcast mode; multicast mode; sensor sharing application; frequency
Autonomous driving is one of the most popular technologies in the automotive industry and even the entire technology industry, and will appear in our lives in the near future. Vehicle-to-vehicle direct communication is essential to support safe and reliable autonomous driving business. The three levels of perception, decision-making and execution of autonomous driving will benefit from the Internet of Vehicles technology and be enhanced. Cellular vehicle-to-everything (C-V2X) technology is widely accepted worldwide as a supplement to other on-board sensors in vehicles, and will be deployed as a leading application of 5G. C-V2X greatly expands the vehicle's ability to detect road participants by providing 360° non-line-of-sight (NLOS) perception. Especially in blind intersections or in bad weather conditions, its advantages over traditional on-board sensors can be better reflected.
In June 2017, the 3rd Generation Partnership Project (3GPP) completed and released the Long Term Evolution (LTE)-V2X R14 standard. R14 LTE-V2X can support basic automotive safety applications. For vehicles, with these communication requirements, they can reliably exchange status information such as location, speed and heading with other nearby vehicles, infrastructure nodes (roadside units) and pedestrians, and can also promptly disseminate warning messages to nearby entities. The spectrum requirements of LTE-V2X have been fully studied by global standardization organizations such as the China Communications Standards Association (CCSA) and the 5G Automobile Association (5GAA), and the consensus is that 20~30 MHz needs to be allocated for LTE-V2X to support basic safety applications, including vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I) and vehicle-to-pedestrian (V2P) applications [1-2] .
3GPP R16 conducted research on C-V2X speech technology and developed the 5G NR-V2X standard based on the 5G New Radio (NR) framework. Its flexible design can support advanced vehicle networking applications that require low latency and high reliability. The flexibility of the NR-V2X PC5 (direct communication) framework allows easy expansion of the NR system to support the further development of more advanced V2X services and other services in the future [3] . 3GPP plans to complete the 5G NR V2X core standardization work in March 2020. 5G NR-V2X technology can further realize and enhance multi-dimensional automation, such as perception, planning, positioning, intent sharing (ADAS) and sensor information sharing. 5G NR-V2X PC5 has different communication modes, including broadcast mode and multicast mode. The multicast communication mode is used to support specific interactive messages within the group, which usually have higher reliability requirements, such as group negotiation, group decision-making, feedback messages, etc. 5G NR-V2X will coexist with LTE-V2X and target different use cases. LTE-V2X will provide basic safety services, while 5G NR-V2X will be used to support advanced automotive applications such as autonomous driving. In this article, we focus on the spectrum requirements for 5G NR-V2X direct communication for autonomous driving.
1. Frequency requirement research method
The relationship between system load and system throughput is used to map the expected packet traffic to the required system capacity, which is called the traffic load mapping method. This method is widely adopted in the capacity and spectrum analysis of V2X direct communication systems [1-2], [4-7] .
For advanced applications in 5G NR-V2X, a new parameter activation factor needs to be introduced, which will reflect the proportion of vehicles sending advanced application messages such as sensor sharing information among all traffic participating vehicles. The spectrum requirement estimate S of 5G NR-V2X can be expressed by formula (1):
(1)
in,
n = 1,..., NbVehiclesInRange. NbVehiclesInRange is the number of vehicles within the effective communication range, which depends on the average speed of the vehicles and the effective communication range.
PS n is the size of the data packet sent by the nth vehicle within the effective communication range, determined by the application traffic model, in bits.
· Ft xn is the frequency of messages sent by the nth vehicle within the effective communication range, determined by the applied traffic model, in Hertz.
SE is the spectral efficiency of the wireless technology, measured in bit/(s·Hz). It is measured at the transmitter side and is determined by the modulation and channel coding scheme used by NR-V2X.
CU is the maximum resource utilization of the wireless channel, reflecting the decrease in spectrum efficiency at the receiving end due to factors such as signal attenuation and co-channel interference.
· DR n is the data rate of the nth vehicle within the effective communication range and can be calculated by PS n ×Ft xn .
· AF AdvApp is the proportion of users who send advanced application messages among all vehicles participating in the traffic, which is called the activation factor.
3GPP has not yet completed the standardization of the radio access part of 5G NR-V2X PC5, so it is necessary to estimate the spectrum efficiency and channel utilization of NR PC5. The spectrum efficiency of NR-V2X can be estimated based on the total number of information bits within a 1 s period and within a 40 MHz channel bandwidth [8] , which is approximately 0.712 bit/(s·Hz).
The channel utilization can be given an assumed range, for example, the lower bound can be the same as the lower bound adopted by the LTE-V2X spectrum demand rate, i.e. 0.336 [9] . For the upper bound, we can assume 80%, which is possible for unicast or multicast communications where better scheduling coordination can be achieved. As 3GPP standardization progresses, more accurate values can be estimated based on system design.
2. Communication requirements and business modeling
In 2016, 3GPP completed the research on communication requirements for 5GNR-V2X [10-12] . Some of the communication requirements were relatively ideal, mainly for the long-term design of autonomous driving systems. Since 2016, autonomous driving technology has developed rapidly, and more practical communication requirements have been proposed for the short-term and medium-term communication requirements. Organizations such as 5GAA have worked closely with automobile manufacturers to refine and summarize the communication requirements [13-14] .
5G NR-V2X PC5 has different communication modes, including broadcast mode and multicast mode. For example, platooning can use multicast mode, while sensor sharing relies more on broadcast mode.
2.1 Broadcast Mode
Sensor sharing, also known as cooperative environmental awareness, is the earliest application to support autonomous driving, which uses the broadcast mode of NR-V2X direct communication.
According to Toyota Motor Corporation [15] , the sensor sharing message size is modeled as 350 B + x * 50 B, where 350 B is assumed to be the average payload size of the basic safety message and x represents the number of other objects observed by the vehicle from the local sensor. 50 bytes represent the amount of information describing a single object. If x is assumed to be 25, then the central red car will perceive the 25 surrounding yellow cars as shown in Figure 1. The autonomous driving message size is 1,250 B and the transmission frequency is 10 Hz [16] .
Figure 1. Surrounding objects observed by local sensors in autonomous driving
For sensor sharing information, if each vehicle sends information about the detected surrounding objects, from the system perspective, a lot of redundant information will be sent. In order to reduce the sending of redundant information, only a part of the vehicles need to send sensor sharing information to share this data with the surrounding vehicles. The ratio of vehicles sending sensor sharing information should be an important assumption to be considered in spectrum demand studies.
The European Telecommunications Standards Institute (ETSI) has developed technical reports and specifications for collective sensing services [17-18] . The ESTI Collaborative Sensing Message (CPM) traffic model is modeled. Due to the different number of surrounding sensing objects, the size of the CPM packet is approximately 550~1900 B [16] .
The following is another example of partially automated driving using information sharing applications given by the 3GPP technical specification. The payload size of the 3GPP traffic model for partially automated driving is 6500 B, which corresponds to the message size of the V2V advanced driving use case [R 5.3-002] in 3GPP TS22.186 [12] . According to the literature [11-12], the message size is assumed to be 60 B, and the payload includes information about 100 objects. Perceiving 100 objects requires very powerful sensor capabilities, and there will be a lot of information redundancy in the air interface. In this case, the payload size of sensor sharing is 6000 B, and another 500 B will be used for coarse driving intention sharing.
2.2 Multicast Mode
Multicast communication is one of the most important functions of 5G NR-V2X. This mode is used to support specific interactive messages within the group, which usually requires high reliability, such as group negotiation, group decision, feedback messages, etc. Therefore, the multicast mode of NR-V2X introduces hybrid automatic repeat request (H-ARQ) [19] to ensure the high reliability and low latency required by group communication. 5GAA has conducted research on use cases and requirements including some group communications [13-14] . Referring to 5GAA's research on group communication use cases and their message flows [13] , we propose a general message flow for multicast, as shown in Figure 2. Figure 2 reflects the general process of multicast process and interaction in group communication, in which general message flow modeling is the basis for business modeling and spectrum requirement research.
Figure 2. Schematic diagram of conventional multicast message flow
3. Evaluation results
3.1 Broadcast Mode
In Section 2.1, we summarized three business modeling schemes for collaborative sensing applications in broadcast mode, namely Toyota research scheme (hereinafter numbered S1), ETSI research scheme (hereinafter numbered S2) and 3GPP research scheme (hereinafter numbered S3).
Table 1 summarizes the business model parameters for collaborative sensing (sensor sharing).
We take a vehicle speed of 70 km/h as an example to calculate the spectrum requirements for sensor sharing. Table 2 gives a summary of the key parameters used for the calculation.
Table 2. Key parameters shared by broadcast mode sensors
For different traffic models S1, S2 and S3, we calculated the spectrum demand and activation factor. The calculation results of the spectrum demand are shown in Figure 3.
Figure 3. Spectrum requirements for broadcast mode sensor sharing
In the early stages of autonomous driving, the percentage of vehicles that can transmit detected objects may be low, and cooperative sensing will not consume too much spectrum. For example, 10 MHz spectrum can be used to provide cooperative sensing services. As V2X vehicles improve their ability to detect objects (i.e. other road participants, obstacles), the message payload size will become larger and cannot be transmitted over the LTE-V2X PC5 interface. Based on the above research, 30~40 MHz of new spectrum needs to be allocated for NR-V2X to carry sensor sharing messages.
3.2 Multicast Mode
The spectrum requirement study method based on the multicast use case is to first estimate the total traffic generated by each vehicle on the multicast use case, and then calculate the amount of spectrum required to accommodate all vehicles within a specified range. Given that the basic principle of multicast is also a physical layer broadcast, and messages are transmitted at maximum transmit power like broadcast, we can use the spectrum requirement study method in Section 2.1. However, it is time-consuming to repeatedly analyze and calculate the spectrum requirements for each use case type, so we propose a novel method to study the spectrum requirements based on the multicast use case. Given the commonality of multicast and broadcast, that is, each vehicle sends messages to all nearby vehicles at maximum transmit power and has similar or identical transmission spectrum efficiency, we can compare the ratio of all traffic generated by multicast messages to the basic safety message (BSM) message load (Ratio_G2B). The ratio of the spectrum requirement of multicast to the amount of BSM spectrum obtained in the literature [2] is also Ratio_G2B. In this way, we can quickly obtain the amount of spectrum required for the multicast use case [5], [8] .
The vehicles participating in the group communication are located in a virtual group as shown in Figure 4. In a group, there is a main traffic participant to lead the group communication of the multicast event.
Figure 4. Communication range diagram of multicast communication
Reference [8] studied the total multicast traffic for a multicast use case “lane change event”. The average traffic rate of each group is the ratio of the sum of the bytes of all messages sent in an event to the average lane change interval T_event, as shown in Figure 5. Assuming T_event is 60 s, the ratio of this traffic to BSM traffic is about 1.5%. With the widespread adoption of cooperative driving strategies, we expect to further reduce lane changes to improve road efficiency. If T_event is extended to 300 s, the ratio of total multicast traffic to BSM traffic is 0.3%. Similarly, we can evaluate the ratio of total multicast traffic for lane change events to total sensor sharing traffic, as shown in Figure 6. Assuming that the size of sensor sharing messages is 1250 B and the sending frequency is 10 Hz according to the traffic model of Toyota Motor Research. We can also find that in the actual road environment, the lane change multicast traffic is negligible compared to the shared sensor traffic. In summary, the multicast traffic of lane changes is negligible compared to the broadcast traffic in a real road environment.
Figure 5. Ratio of lane change traffic to BSM traffic
Figure 6. Ratio of lane change traffic to sensor sharing traffic
4. Conclusion
In the NR-V2X system, messages carrying status information and environmental information (such as sensor sharing messages) can be sent in broadcast mode. These messages will consume major spectrum resources, and studies have shown that at least about 30~40 MHz of spectrum is required. NRV2X uses multicast mode to send negotiation and decision messages in group communication. Based on the study of some multicast-based use cases, we noticed that these multicast applications are always event-triggered, and the probability of event occurrence is usually low; therefore, although the multicast mode is more critical to supporting advanced applications, the total traffic volume transmitted through the multicast mode is much less than the total traffic volume of broadcast messages. If the traffic volume of multicast is negligible compared to broadcast messages, the spectrum required for multicast can be ignored relative to the spectrum required for broadcast messages such as BSM and cooperative sensing. In the early stages of NR-V2X frequency research, the frequency requirements of multicast mode can be temporarily ignored. Based on current research, we can draw the following conclusion: at least 40 MHz of spectrum is required to support different sensor fusion, path planning algorithms, and group communications to support the upcoming autonomous driving.
5.9 GHz is the spectrum for intelligent transportation systems (ITS) for global and regional integration of the International Telecommunication Union Radiocommunication Sector (ITU-R) [20] , which can bring economies of scale to the development of C-V2X and related ITS services. In addition to the 20 MHz frequency allocated in 5.9 GHz for LTE-V2X to provide basic safety services, at least 40 MHz should be reserved for 5G NR-V2X direct communication (broadcast mode, multicast mode and unicast mode) to support the autonomous driving technology to be deployed in the near future.
Acknowledgements
This research was supported by the members of the FuTURE&TIAA Internet of Vehicles Joint Working Group. We would like to express our gratitude to them!
References
[1] Study on Frequency Requirements and Related Interference Coexistence of Intelligent Transportation Systems/Vehicle-Road Active Safety Applications: CCSA SR 239-2018 [S]. CCSA, 2018
[2] Study of Spectrum Needs for Safety Related Intelligent Transportation Systems-Day 1 and advanced Use Cases: 5GAA S-190179 [S]. 5GAA, 2019
[3] New SID: Study on NR-V2X: 3GPP RP-181429[S]. 3GPP, 2018
[4] GAO L, LI Y, MISENER J, et al. C-V2X Based Basic Safety Related ITS Spectrum Requirement Analysis[C]//2017 IEEE 86th Vehicular Technology Conference (VTC-Fall). USA: IEEE, 2017
[5] Spectrum Requirements for 5G NR-V2X Direct Communication for Autonomous Driving V2.0[R]. Future Mobile Communications Forum, 2019
[6] Li Yan. 5G and Internet of Vehicles[M]. Beijing: Electronic Industry Press, 2019
[7] 5GAA. Study of Spectrum Needs for Safety Related Intelligent Transportation Systems: 5GAA_S-180026[S]. 5GAA, 2018
[8] 5G NR-V2X PC5 Spectrum Needs Study: Methodology for Groupcast: 5GAA S-190215 [S]. 5GAA, 2019
[9] Study of Spectrum Needs for Safety Related Intelligent Transportation Systems: 5GAA_S-180026[S]. 5GAA, 2018
[10] Service Requirements for V2X Services: 3GPP TS 22.185v14.0.0[S]. 3GPP, 2015
[11] Study on enhancement of 3GPP Support for 5G V2X Services: 3GPP TR22.886V15.0.0[S]. 3GPP, 2016
[12] Enhancement of 3GPP Support for 5G V2X Services; Stage 1: 3GPP TS22.186V15.2.0[S]. 3GPP, 2017
[13] 5G Use Cases and Requirements-Wave 2.1:5GAA TR T-190028[S]. 5GAA, 2019
[14] 5G Use Cases and Requirements-Wave 2.2::5GAA TR T-19099[S]. 5GAA, 2019
[15] BANSAL G. The Role and Design of V2X Communications for Automated Driving[R]. Toyota, 2016
[16] 5G NR-V2X PC5 Spectrum Needs Study: 5GAA S-190017 [S]. 5GAA, 2019
[17] Intelligent Transport Service; Vehicular Communications; Basic Set of Applications; Informative Report for the Collective Perception Service: ETSI TR 103 562 V0.0.14[S]. ETSI, 2018
[18] Intelligent Transport Service; Vehicular Communications; Basic Set of Applications; Specification of the Collective Perception Service: TS 103 324 V<0.0.12>[S]. ETSI, 2017 [19] Physical Layer Procedures for Sidelink: 3GPP R1-1907274 [S]. 3GPP, 2019
[20] Harmonization of Frequency Bands for Intelligent Transport Systems in the Mobile Service [EB/OL]. [2020-12-22]. https://www.itu.int/md/R15-SG05-C-0120
|