Accelerating Collaboration and CV Deployment

New York City CV Pilot to Use High-Accuracy Positioning Techniques

The Connected Vehicle (CV) Pilot project currently being implemented by the New York City Department of Transportation (NYCDOT) will deploy multiple CV safety-related applications in parts of Manhattan and Brooklyn, consistent with the City’s focus and dedication of resources to achieve its Vision Zero goal of reducing traffic-related injuries and fatalities to zero.

The CV Pilot devices on board vehicles and installed in the infrastructure will enable ultra-fast vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication of safety and mobility related messages to enable applications to alert drivers to sudden braking ahead, assist in reducing intersection conflicts and potential red light violations, and support safe vehicular operations such as compliance with vehicle height restrictions and city speed regulations.  V2V applications depend on relative positioning (assessing the position and trajectory of a remote vehicle as reported in transmitted Basic Safety Messages (BSMs) relative to the host vehicle’s location/trajectory), while V2I applications depend on absolute positioning (relating the host vehicle’s position to the absolute position of the infrastructure – such as lanes or physical structures).  The position accuracy required for successful operation of an application may vary – for example knowing the relative (lateral) lane position is important for Forward Collision Warning but not as much for Intersection Movement Assist, where vehicle paths are crossing.

Taxis, city buses, and NYC fleet vehicles will be among the major participants.  Most of these vehicles, as well as approximately 350 intersections with roadside units (RSUs) will be focused on lower Manhattan, where the density of tall buildings (often called an “urban canyon”) is legendary.  Tall buildings tend to obstruct and reflect signals from satellites, complicating the task of providing vehicles with accurate locations using Global Positioning System (GPS) and other global navigation satellite systems (GNSS).

Recognizing these challenges, the NYCDOT team and their suppliers explored various supporting techniques to improve location accuracy, including dead reckoning, CAN bus integration for speed information, inertial measurement unit integration, and road side unit to vehicle ranging with software that uses messages received from multiple RSUs. Working with an Aftermarket Safety Device (ASD) vendor, a hybrid capability was tested and demonstrated on 6th Avenue in Manhattan. The capability utilizes a combination of inputs from vehicle sensors and GNSS signals when available, and standard RSU-transmitted advertisement messages from the NYCDOT CV Pilot-installed equipment to establish the vehicle’s location with lane level accuracy.  Preliminary testing results are very promising and exceed the threshold (within 1.5 m 68% of time) referenced in SAE J2945/1, even under these “urban canyon” conditions.

While this capability does require software on-board the ASD to perform the processing of inputs and data fusion, it does not require additional roadside infrastructure beyond the NYCDOT1 RSUs being installed, broadcasting standard service advertisements. Outside of the NYCDOT CV Pilot-equipped areas, the ASD will need to utilize other positioning strategies, such as GNSS in conjunction with dead reckoning via speed inputs; however, the positioning accuracy can vary depending on location.

Earlier in the project, NYCDOT and several vendors explored the use of Ultrawide Band (UWB) radio technology for positioning enhancements in NYC. This approach would have mounted transponders on top of poles to transmit precise signals to allow high-accuracy location measurement by in vehicle devices. However, for the NYCDOT CV Pilot, the cost of deployment and the additions required to the in-vehicle unit were cost prohibitive, and the proposed solution was not considered ready for production use in the project at this time. Also, NYCDOT originally was expecting to broadcast RTCM (Radio Technical Commission for Maritime Services) positioning corrections from RSUs for use by the ASD-equipped vehicles, but they are no longer planning to do so; the other two CV Pilot sites (Tampa, FL and the state of Wyoming) have also stated that their vehicles are not planning to use RTCM messages. Vehicles from the other sites should also be able to receive the standard RSU advertisement messages, but without the RSU time-of-flight measurement capability, vehicles for these sites are not expected to perform well in New York’s urban canyons.  Initial testing for site interoperability occurred under “open sky” conditions, i.e. in locations without tall buildings or other reflective objects, and did not incorporate the RSU time-of-flight feature.

It should be noted that as the state of positioning technology evolves over time, the capabilities, costs, and availability of solutions may change. Research is ongoing for improving both standalone and augmented positioning performance in urban environments. Currently, there are significant tradeoffs in cost versus performance for standalone systems, as the cost of more capable positioning systems is currently outside the range of the automotive market. However, this has potential to change as the market and technologies change with automation and satellite constellation enhancements. For urban canyon environments, results from the NYCDOT CV Pilot experience should be of great benefit in informing other cities with similar dense grid street networks of the potential for enhanced positioning utilizing RSUs at intersections. This experience should also help to understand positioning-related topics in CV, such as positional accuracy for the BSM.


NYCDOT CV Pilot Update at the System Design Milestone (2017). Slides available at

NYCDOT CV Pilot Project, presentation at 25th ITS World Congress (2018). Slides available at

1. From NYCDOT RSU specification: RSUs shall exceed 802.11 ACK requirements in the following manner: Antenna referenced ACK turnaround time must be within SIFS +/- 12.5 ns (95% of time) for cable tested non-CSD signals.  Note that any RSU employing the NXP based SAF5200 will support this by default.  The RSU position provided by the WSA shall be provided by the central system based on the 3D surveyed position.