Connected vehicles, which are likely to be an important component of smart cities, require spectrum. As of December 2019, the US Federal Communications Commission and the US Department of Transportation were pursuing opposing visions of a spectrum policy for connected vehicles. Through extensive simulation, this project will provide both agencies with objective analysis to make informed decisions about how much spectrum to allocate for DSRC and/or C-V2X, and how to channelize the spectrum.
CONTEXT AND OBJECTIVES
Two decades ago, the U.S. Federal Communications Commission (FCC) established an Intelligent Transportation System (ITS) band of spectrum for connected vehicles. It would support wireless communications from vehicle to vehicle (V2V), vehicle to infrastructure (V2I), and vehicle to pedestrians, cyclists, and everything else (collectively called V2X). Very recently, that once-stable decision has given way to chaos and uncertainty.
The FCC initially allocated 75 MHz of spectrum, which was available for connected vehicles on an exclusive basis, and they limited use to a single V2X technology known as Dedicated Short-Range Communications (DSRC). This 75 MHz was to be used for both safety-critical applications, and non-safety transportation applications that may be of interest to many communities, especially in smart cities [PEHA17c, PEHA18d]. If that decision holds, DSRC will be an important part of smart city infrastructure. Some cities have already deployed some DSRC infrastructure on a limited basis, e.g. Pittsburgh uses it for smart streetlights. However, usage has been modest to date. In recent years, there has been intense debate at the FCC over whether to make some of the 75 MHz ITS band available to unlicensed devices unrelated to transportation, such as Wi-Fi, on a shared basis. Two methods of sharing spectrum between connected vehicles and unlicensed devices have gained prominence, and testing is currently underway to determine whether either or both would be sufficiently safe. In the meantime, a technology called C-V2X has emerged from the cellular industry to rival DSRC, although C-V2X is still prohibited in the ITS band. Some companies had urged the FCC to make some of the ITS band usable for C-V2X. That was the state of the debate until December of 2019, when things changed dramatically.
On December 10, the U.S. Department of Transportation (DOT) held a public meeting to make the case that, for the sake of roadway safety, it is imperative that connected vehicles have access to the full 75 MHz of spectrum, and that none of it be shared with non-transportation systems. DOT did not take a position on whether DSRC or C-V2X was best, which may be an implicit (although perhaps not deliberate) endorsement of DSRC, just because DSRC is the status quo. A mere two days later, the FCC commissioners voted to consider alternative policies, where ITS would tentatively lose 60% of its spectrum. A set of possible new rules were released last week [FCC19]. Of the 30 MHz remaining, either 20 MHz would go to C-V2X and 10 MHz to DSRC, or all 30 MHz would go to C-V2X and DSRC technology would be prohibited entirely in the ITS band. Some Commissioners said that the ITS band should be used only for safety-critical applications, contradicting previous FCC and DOT positions. In the public meeting, there was no mention of the fact that the FCC has not even completed testing of the last spectrum policy that it was considering, which involved sharing the ITS band. Thus, two federal agencies are now openly pushing in opposite directions with respect to connected vehicles, and other government agencies that should also play a role have so far remained silent. Both the FCC and DOT are lacking good quantitative evidence to show that they are pushing in the right direction.
In the proposed research, we will investigate some of these spectrum issues. In particular, we will examine how much spectrum is needed for the safety-critical applications alone, and how this differs with DSRC versus C-V2X technology. Is it the full 75 MHz? Is it the 30 MHz that would remain in the ITS band if the latest proposal is adopted? Is it 10 MHz, which is all that would be available for DSRC under one of the proposed plans?
The issue is complicated by the fact that the FCC’s spectrum decisions must meet connected vehicle needs for many years to come, even though the applications that must operate over that spectrum are now in flux. Our understanding of V2X applications has certainly changed since the 75 MHz was allocated, but it would also be a mistake to allocate spectrum specifically for 2019 applications when autonomous vehicles are on the horizon. Thus, in this research we will explore both current and potential future applications. If there is enough spectrum in the ITS band, the FCC may also require that different types of network traffic be carried in different channels, so that the most critical traffic flows on the less congested channel. We will explore that as well.
This builds upon years of research we have undertaken on connected vehicle networks, which has included collecting extensive data from a citywide deployment of DSRC, using that data to model vehicular movement and wireless communications, and creating extensive simulations of V2X systems [PEHA15, PEHA16a, PEHA16b, PEHA17a, PEHA17b, PEHA18a, PEHA18b, PEHA18c, PEHA19].
To see the impact of different decisions about spectrum, including the amount of spectrum allocated to ITS, the technology choice, and the channelization rules, we will extensively simulate vehicular networks. We will do so under different spectrum management approaches, different technology choices, different population densities from dense urban to sparse rural, different applications and different scenarios. To do that, we need realistic models of vehicle mobility behavior, of signal propagation, of network protocols, and of applications.
The first step is to characterize vehicle mobility. In past research, we have collected data on the mobility of many hundreds of DSRC-equipped cars, which includes location and velocity readings every 1 to 5 seconds over a period of months to years, and used that to simulate mobility. We now supplement that with data obtained from the Department of Transportation of other trials [DOT19a], such as the one in Ann Arbor [DOT19b[. The DOT data is less detailed, but includes thousands of vehicles, and a more diverse set of locations. From this data, we can construct realistic models of how the geographic concentration of vehicles changes over time, since network congestion and interference come from such concentration, and how vehicles accelerate and decelerate as they approach intersections of hazards, which is critical for assessing safety applications.
The next step is to characterize the network traffic generated within a single vehicle or V2X-equipped piece of infrastructure by applications of interest. We first survey the current and proposed applications that enhance safety by providing real-time guidance to human drivers. For example, applications warn drivers when a bridge is frozen, when a signal light is about to change, or when it is unsafe to turn left at an intersection due to oncoming traffic. After years of technical progress, dozens of such applications have been defined [NHTS14, DOT16, AREN19, MIUC18, WANG17]. We’ll produce a detailed characterization of the network traffic that applications generate so we can determine the impact on other vehicles, and a high-level understanding of the performance required by the application. (The latter will be investigated more thoroughly in later stages of the research.)
We then similarly survey and characterize applications that are likely to be important for use in partially or completely autonomous vehicles, which is a newer area of research. For these applications, we envision vehicles that are under the control of computers, at least temporarily, and that use information obtained from V2X communications to complement information obtained via sensors. For example, some applications will control cars only as they go through a given intersection. We will examine the range of applications that have been proposed or considered, so we can model each and every packet they transmit over ITS spectrum.
Among the most important and the most technically challenging examples is platooning. In platooning, multiple vehicles travel in succession, and except for the platoon leader, each vehicle follows the vehicle in front of it using autonomous means. It has been demonstrated that platooning can be achieved using sensors and no V2X communications, but the distances between vehicles can be decreased with V2V communications [MIKA19]. Decreasing the distance between vehicles means longer platoons, better fuel efficiency, and ultimately lower costs. The application is particularly challenging because there is no way to prevent multiple large platoons from converging in the same place and time and thus sharing spectrum, and because poor network performance can contribute to fatal accidents. It is therefore necessary to design for the worst case. We will simulate platooning at length, while varying number of lanes per highway, number of vehicles per platoon, velocity, and other parameters.
All of this feeds into detailed packet-level simulations. We have developed extensive software for the simulation of DSRC networks, and have begun developing software for the simulation of C-V2X in more limited cases. We will continue to expand.
We assume that ITS spectrum is divided into 10 MHz channels. We will vary the number of channels, and their purpose. Each channel may be for DSRC or C-V2X (mode 4), for different types of applications (e.g. beaconing, platooning, diagnostics, etc.). We will determine which spectrum strategies can protect safety-critical traffic over a wide range of applications and conditions.
We will initially quantify performance results by examining load at every location and in every channel in any given scenario, where load at a location and channel is defined as the expected value of the number of transmissions underway simultaneously that are capable of causing significant interference at that location. Where load is low, there is little risk to safety-critical communications. Where load is high, one must delve more deeply. This measure does depend on physical-layer properties such as transmission power and antenna height, but not on protocol, so it will be particularly useful in allowing the FCC to reach spectrum management decisions that can be separated from the fight over DSRC vs. C-V2X.
The next level of detail for performance results is to determine network-level measures of quality of service, including packet loss, latency, and throughput, all as a function of distance. Other researchers have compared technologies using these metrics, but there is much to do in examining the impact of the spectrum management decisions discussed above.
Ultimately, we hope to examine application-layer performance rather than network-layer performance, as it is application-layer performance that really determines safety. For each application, we must employ the right performance metrics. In many (although not all) safety applications, vehicles and infrastructure send out information about current status, such as a car’s position and speed, or the extent to which a bridge has frozen over. For such messages, packet loss probability, latency, and throughput are of little value. Since each message from a given source makes all previous messages irrelevant, what really matters at a given time is the age of the last message received from that source. For such applications, we break with the conventional approach by focusing on the mean, 95th percentile, and distribution of this age, rather than something as meaningless as packet loss. Whether age grows large because a device is not able to access the channel or because packets are transmitted but lost due to poor SINR makes no difference, and the fraction of packets lost matters far less than the number of losses that are consecutive. (The distinction matters most in C-V2X, which supports periodic timeslot assignments, since there are ways one might accept high loss rates in even-numbered packets but not in odd-numbered.) For other applications, we will similarly choose performance metrics that matter at the application layer, and quantitatively assess them in various scenarios.
As one example, we have begun analysis of one particular application known as Forward Collision Warning (FCW) [KUSA12, DOT16]. Every 100 ms, vehicles send out messages with current location, velocity and acceleration on a specific 10 MHz channel. In the trailing vehicle, an application receives those messages from the leading vehicle, estimates P, which is the probability that a driver needs to brake hard to avoid a collision [KIEF03, YASE16]. The model was derived by observing reactions of human drivers. The application warns the driver if estimated P exceeds some threshold. Transportation researchers typically assume that the following vehicle always has current information about the lead vehicle, but in reality, poor performance by the V2X communications system can lead to poor estimates of P. Through simulation, we are determining how various factors affect the accuracy of estimated P. In the proposed research we can consider how spectrum is managed as a critical factor.
Another interesting example of an application worth exploring is platooning. Other researchers have determined appropriate control mechanisms for the i’th truck in a platoon as a function of the distances, velocities and accelerations of the truck(s) in front, e.g. [RAJA98, RAJA12, VUKA18]. They typically simulate the behavior of all trucks in a platoon with a given control mechanism to make sure that the risk of collision of sufficiently low. We can do similarly, except in our simulations, information about distances, velocities and accelerations can be outdated, because of the time it takes to successfully send that information via DSRC or C-V2X, With our simulations, we can see how network conditions affect safety, and how spectrum management decisions affect both network conditions and safety.
In conclusion, by seeing how spectrum management decisions affect load at each location, network-layer performance for each pair of devices that communicate, and application-layer performance for a variety of important applications, we can help the FCC make an informed decision about this critical issue.
When we have useful results, they will be presented directly to the FCC, and perhaps the Department of Transportation as well. The PI of this proposal is a former Chief Technologist of the FCC, and advises there on a regular basis. He has advised senior staff at the FCC on connected vehicles, including a personal briefing to the Chairman of the FCC.
[AREN19] Arena, F., & Pau, G. (2019). An overview of vehicular communications. Future Internet, 11(2), 27.
[DOT16] U.S. Department of Transportation, “Federal Motor Vehicle Safety Standards; V2V Communications - Notice of Proposed Rulemaking (NPRM)” (2016).
[DOT19a] US Department of Transportation, Public Data Portal, Automobile Data and Resources, https://data.transportation.gov/browse?category=Automobiles
[DOT19b] US Department of Transportation, Safety Pilot Model Deployment Data, https://data.transportation.gov/Automobiles/Safety-Pilot-Model-Deployment-Data/a7qq-9vfe
[FCC19] Federal Communications Commission, Notice of Proposed Rulemaking, in the Matter of Use of the 5.850-5.925 GHz Band, ET Docket No. 19-138, December 17, 2019.
[KIEF03] Kiefer, R. J., Cassar, M. T., Flannagan, C. A., LeBlanc, D. J., Palmer, M. D., Deering, R. K., & Shulman, M. A. (2003). Forward collision warning requirements project: refining the CAMP crash alert timing approach by examining" last second" braking and lane change maneuvers under various kinematic conditions (No. DOT HS 809 574). United States. National Highway Traffic Safety Administration.
[KUSA12] Kusano, K. D., & Gabler, H. C. (2012). Safety benefits of forward collision warning, brake assist, and autonomous braking systems in rear-end collisions. IEEE Transactions on Intelligent Transportation Systems, 13(4), 1546-1555.
[MIKA19] Mikami, M., & Yoshino, H. (2019). Field Trial on 5G Low Latency Radio Communication System towards Application to Truck Platooning. IEICE Transactions on Communications
[MIUC18] Miucic, R. (Ed.). (2018). Connected Vehicles: Intelligent Transportation Systems. Springer.
[NHTS14] U.S. National Highway Traffic Safety Administration, Vehicle-to-Vehicle Communications: Readiness of V2V Technology for Application, August 2014.
[PEHA15] Alexandre Ligo, Jon M. Peha, Pedro Ferreira and Joao Barros, "Comparison between Benefits and Costs of Offload of Mobile Internet Traffic Via Vehicular Networks," Proceedings of 43rd Telecommunications Policy Research Conference (TPRC), Arlington, VA, September 2015.
[PEHA16a] A. Ligo, J. M. Peha and Joao Barros, "Throughput and Cost-Effectiveness of Vehicular Mesh Networks for Internet Access," IEEE 84th Vehicular Technology Conference (VTC), Sept. 2016.
[PEHA16b] A. Ligo and J. M. Peha, “Cost-Effectiveness of Using Connected Vehicles Infrastructure for Internet Access,” MASITE/ITSPA Annual Conference, 2016.
[PEHA17a] A. Ligo and J. M. Peha, "Is It Cost-Effective to Share Roadside Infrastructure for Non-Safety Use?," IEEE 85th Vehicular Technology Conference (VTC), June 2017.
[PEHA17b] A. Ligo and J. M. Peha, "Spectrum Policies for Intelligent Transportation Systems," 45th Telecommunications Policy Research Conference (TPRC), Sept. 2017.
[PEHA17c] J. M. Peha, KEYNOTE: “Wireless Communication and Municipal Governments – Looking Forward,” Association of Boroughs, August 2017.
[PEHA18a] A. Ligo, J. M. Peha, Pedro Ferreira and Joao Barros, "Throughput and Economics of DSRC-Based Internet of Vehicles," IEEE Access, vol. 6, pp. 7276–90, 2018.
[PEHA18b] A. Ligo and J. M. Peha, "Cost-Effectiveness of Sharing Roadside Infrastructure for Internet of Vehicles," IEEE Transactions on Intelligent Transportation Systems, Volume 19, Issue 7, July 2018, pp. 2362-2372.
[PEHA18c] A. Ligo and J. M. Peha, "Spectrum for Intelligent Transportation Systems: Allocation and Sharing," IEEE International Symposium on Dynamic Spectrum Access Networks (DySPAN), Oct. 2018.
[PEHA18d] J. M. Peha, “Smart City Technologies for Local Governments,” Fall Conference of Townships, Boroughs & Authorities, August 2017.
[PEHA19] Alexandre Ligo and Jon M. Peha, "Spectrum for V2X: Allocation and Sharing," accepted to appear in IEEE Transactions on Cognitive Communications and Networking.
[RAJA98] Rajamani, R., Choi, S. B., Law, B. K., Hedrick, J. K., Prohaska, R., & Kretz, P. (1998). Design and experimental implementation of longitudinal control for a platoon of automated vehicles. J. Dyn. Sys., Meas., Control, 122(3), 470-476.
[RAJA12] Rajamani, R. (2012). Vehicle Dynamics and Control Springer New York.
[VUKA18] Vukadinovic, Vladimir, Krzysztof Bakowski, Patrick Marsch, Ian Dexter Garcia, Hua Xu, Michal Sybis, Pawel Sroka, Krzysztof Wesolowski, David Lister, and Ilaria Thibault. "3GPP C-V2X and IEEE 802.11 p for Vehicle-to-Vehicle communications in highway platooning scenarios." Ad Hoc Networks 74 (2018): 17-29.
[WANG17] Wang, X., Mao, S., & Gong, M. X. (2017). An overview of 3GPP cellular vehicle-to-everything standards. GetMobile: Mobile Computing and Communications, 21(3), 19-25.
[YASE16] Yaser P. Fallah and Masoumeh K. Khandani, “Context and Network Aware Communication Strategies for Connected Vehicle Safety Applications,” IEEE Intelligent Transportation Systems Magazine 8, no. 4 (2016): 92–101, https://doi.org/10.1109/MITS.2016.2593672.
In this project, deployment means that government policymakers will make decisions using our results. The PI expects to brief policymakers at the FCC and elsewhere as appropriate throughout the project. The Federal Communications Commission (FCC) has announced that it will undertake a proceeding to consider new rules on this topic in 2020, which many believe will end around January 2021. Other proceedings of the FCC are likely to follow, as well as the Department of Transportation.
Expected Accomplishments and Metrics
||Carnegie Mellon University
||Faculty - Tenured
||Carnegie Mellon University
||Student - PhD
Amount of UTC Funds Awarded
Total Project Budget (from all funding sources)
|Data Management Plan
||Feb. 8, 2020, 7:23 p.m.
||Feb. 9, 2020, 7:17 a.m.
||Leading the Way: A National Task Force on Connected Vehicles
||Sept. 10, 2020, 6:19 p.m.
||KEYNOTE: Spectrum Policy for Intelligent Transportation Systems
||Sept. 11, 2020, 7:43 a.m.
||Improving Spectrum Efficiency for 5G through Multi-Network Access
||Sept. 11, 2020, 7:43 a.m.
||Emerging Wireless Technology and Competition
||Sept. 11, 2020, 7:43 a.m.
||Technical Standards and Spectrum Sharing for Intelligent Transportation Systems
||Sept. 10, 2020, 6:32 p.m.
||The Need for Analysis and a Shared Vision for Intelligent Transportation Systems
||Sept. 10, 2020, 6:32 p.m.
||Sept. 23, 2020, 6:12 p.m.
No match sources!
|City of Pittsburgh