Simulating Connected Vehicles with the V2X Framework in Aimsun Next
21 August 2018
Simulation with Aimsun Next software plays a key role supporting the FLOURISH project by providing a set of tools for investigation into the use of Vehicle-to-Everything (V2X) connectivity. The new, extensible V2X Framework in Aimsun Next can help researchers design innovative new systems based around connected vehicles and allow them to test future urban traffic management systems, in-car information tools, and autonomous vehicle controls. These systems will be able to use the rich source of data from connected vehicles to help manage road traffic and soon will be able to provide more sustainable transport and a better journey experience.
The challenge is that this needs to take into account many unknowns: the different technologies (5G phones, WiFi, ZigBee…) used to transmit messages; the different parameters of each technology for throughput, reliability and range; the different types of messages that vehicles may send (only a few of them are standardised); and whether vehicles are conventional cars with a driver or the connected and autonomous vehicles of the future. The way these systems and vehicles use information to respond to what they have learned from the V2X communications will be very different.
This is why the communications extension to Aimsun Next, developed within the FLOURISH project, is called a V2X Framework and has been designed to be readily extended to add new technologies and messages as they emerge. It means that we can’t assume there is a small and simple set of messages and communications hardware that we can simulate to do everything we need. Instead, what is required is an extensible software environment that we can build on to add the new technologies and message types as they emerge; an environment with the flexibility to work with vehicle behaviour that ranges from detailed collaboration between connected and autonomous vehicles, to showing a driver information in the hope of eliciting a reaction.
To illustrate the need for flexibility, there are just a few standardised message protocols; i.e. the CAM (Collaborative Awareness Message) protocol, which defines the data that a vehicle can broadcast about itself. However, a quick look at the ‘Connected Vehicle Reference Implementation Architecture’ ( https://local.iteris.com/cvria/), shows that there are already over 100 potential message types proposed, which means it’s not possible for Aimsun software to automatically include them all, even if that were the full list. In fact, by default, the V2X Framework includes only a small number of the existing EU and American defined types, which means that, for new experimental CAV applications, the modeller needs to extend a basic message class to make a new message type. The V2X Framework provides the software systems to do this and to have the new message slot seamlessly into the system.
When we look at simulating how the messages are transmitted (as opposed to the data the messages contain), the V2X Framework, by default, has a simple model of a radio with just a transmission range and a probability of successful transmission. This is obviously very quick to simulate but, in some cases, more accurate representation of the radio waves will be needed. Therefore, in the same way as we can add new messages, we can also program new transmitters and add them to the V2X Framework.
In summary: the V2X Application Programming Interface (API) in Aimsun Next is a software development toolkit that is designed to be flexible and extendable so that a researcher can add to it to include new messages and radio technologies with minimum effort and with no need to reconfigure the whole system.
So, how does a modeller or system designer go about using this toolkit? What the modeller needs to do is to write some code, both for the vehicle and for the Road Side Units (RSU), which, at every time step in the simulation, goes though the list of messages received by each vehicle or RSU, takes the data from them, does something with that data, and finally transmits any new messages required.
Evaluating what the vehicle, the driver, or the network manager, will do with that data is actually the more complex task. Interesting questions arise for the modeller about how to communicate information to drivers to ask them to change what they are doing and then to evaluate what they will actually do. Some drivers will trust the information, some ignore it, and some interpret it, thinking that they know better. Then there is the related, but different, question about the effect of 100 connected and autonomous vehicles getting the same information and those 100 controllers all deciding to react in more or less the same way as they all use the same algorithms; the effect on the road network may be not what was planned. This is why we want to use simulation, to observe and understand these problems in a virtual environment before encountering them in reality.
Also, from the modeller’s point of view, the simulation also needs to operate at different scales. In one scenario, we may be testing how vehicles can collaborate with each other, to drive fast through a junction, missing each other by just enough, and no more. That needs a very high accuracy in the transmitter simulation because that is the critical component of the vehicle movement. However, simulating at this level of detail across a wide-area network would slow the computer to a crawl. Then, in another scenario, we may want to test how we aggregate information from many vehicles to understand more about congestion and how to advise a driver to change route to avoid it. This needs a simulation that covers a wide area with many alternate routes in a complex road network and in this case, the modeller is more concerned with how many vehicles are contributing to the information and so the transmitter simulation can be simplified, allowing this wide area simulation to run at a reasonable speed.
Modelling at an appropriate scale, from a single junction to a city-wide road network, is an important part of what Aimsun Next is able to do with integrated macro, meso and micro models. The design of the V2X Framework in Aimsun Next, with its inherent extensibility and flexibility, is just a part of the same philosophy.
Authored by: Aimsun
Back to List