The deployment of Connected and Autonomous Vehicles (CAVs) will change the way we live. In particular, Connected Vehicles will allow us to build an Intelligent Transportation System (ITS) in which there is strong cooperation between vehicles and the transport infrastructure. This is referred to as Cooperative-ITS (C-ITS). The deployment of C-ITS will lead to better traffic and road management, shorter journey times, less accidents, better collision avoidance mechanisms and increased efficiency in the management of major disasters.
In order to understand this coming age, it is necessary to build new technologies, testbeds and applications that will give us insight into this brave new world. The Department for Transport (DfT) and Middlesex University (MDX) have built a Connected Vehicle Testbed using ETSI ITS-G5 (also known as VANET) technology. The testbed has been built on the Hendon Campus in London and surroundings roads and also extends along the A41 (Watford Way) behind the campus. The testbed is now fully operational and trials have begun to fully understand the technology and issues around its wide-scale deployment as well as the building of an Intelligent Information Platform (IIP) to help create real applications for this environment. This report looks at the design and implementation of the testbed and how this effort has uncovered key challenges in deploying C-ITS in the UK.
1.0 Introduction to the Technology
Vehicular Ad-Hoc Networks (VANETs/ETSI ITS-G5) are new networks that will enable support for life-critical, safety and infotainment applications. These networks are deployed using Roadside Units (RSUs) that are placed along the road and OnBoard Units (OBUs) that are placed in vehicles or worn by cyclists and pedestrians. OBUs can communicate with RSU and with each other using DSRC/IEEE 802.11p protocols, which provide low latency and high bandwidth. Examples of VANET/ETSI ITS-G5 communication are shown in Figure 1. This technology has been standardised to operate at 5.9 GHz and has a coverage range of around 1km.
2.0 The Middlesex Research Testbed
In August 2014, Middlesex University decided to explore the building of a testbed to study seamless communication in highly mobile environments at its Hendon Campus. It was decided to use VANET Technology because of its coverage range and its ability to support a set of diverse applications. 4 RSUs and 10 OBUs were purchased. Photos of an RSU and an OBU are shown in Figure 2.
The four RSUs were mounted on four buildings on the campus and they were able to provide access to vehicles within the campus and the surroundings roads. Figure 3 shows the system setup.
Readings from the OBUs in the cars are sent to the RSUs and the data is sent to a WSMP Server program, which ran on an MDX Server machine. The data on the server can be accessed by a client program application, which then displayed the data using Google Earth.
In June 2015, Middlesex University won a Transport Technology Research Innovation Grant (T-TRIG) award. This allowed us to run a Middlesex VANET Trial in January 2016. During this trial, OBUs were placed in the cars of students and staff. The readings for a single day in January 2016 are shown in Figure 4.
The results in Figure 4 show the readings obtained from each RSU as cars were driven around the campus and surrounding roads. The figure clearly shows readings being obtained along the A41. For the red dots on the A41, these readings were obtained by the RSU on the Williams building which as between 1.1 and 1.2 kms from where these readings were recorded and thus exceeded the limit of the standard coverage range. This result was due to the fact that the Hendon Campus is on a hill and therefore there was a good line of sight between vehicles on the A41 and the Williams RSU.
3.0 Extending the Testbed
Given these results, it was decided to extend the testbed by putting RSUs directly onto the A41. This was undertaken in late November 2016 and three additional RSUs were mounted along the A41. Transport for London (TfL) provided the manpower and equipment to help us get this done. However, since the RSUs were being placed outside the campus, data from these RSUs could not be backhauled using the University network as was previously done. It was decided to use LTE (4G) to backhaul the motorway RSUs. Machine-to-machine (M2M) SIMs were provided by Mobius Networks to backhaul the data from the motorway RSUs to the MDX Cloud Server.
This work has recently been completed and a new coverage map has been produced. This is shown in Figure 5. All seven RSUs are fully operational. The first four (1-4) RSUs cover the Hendon Campus and surrounding roads. This covers an area of around 0.7 miles or 1.1 kms. The A41 is covered by the other (5-7) RSUs. The coverage runs from the entrance of the Great Northern Way (top of brown line) to Hendon Central Tube Station (bottom of the blue line). This is a distance of 2 miles or 3.2 kms. Hence the total coverage of the testbed is 2.7 miles or 4.31 kms. This Middlesex testbed is interesting because it encompasses motorway as well as urban roads and thus can allow a comprehensive study of the traffic congestion caused by the interaction of these two environments.
4.0 Challenges for C-ITS
Building a fully functional testbed has highlighted some key issues that need to be addressed in order to facilitate the large-scale deployment of C-ITS.
4.1 Wide-Scale Deployment of C-ITS
A major issue is to find the optimum deployment of RSUs because, in a wide-scale deployment of this technology, the cost of the RSUs will likely be the most expensive per unit CAPEX expenditure. It is therefore necessary to be able to keep this cost to a minimum.
However, in order to achieve this, it is essential to get very accurate propagation models to estimate the best locations for the RSUs. As we see with the testbed there are options with regard to where the RSU should be placed: either on the lampposts on motorways, on tall buildings in urban areas or on mobile phone masts. With tall buildings, the Free Space Path Model (FSPM) can be used, however this model cannot be used for mobile phone masts or lampposts. In addition, the coverage maps of the mobile phone masts, held by the mobile operators, are based on the operating frequencies for mobile phones and not the 5.9 GHz operating frequency of VANET/ETSI ITS-G5 technology. Furthermore, as we have seen with the testbed, the coverage area of a given RSU on a lamppost is very much dependent on potential obstacles as well as the topology of the road infrastructure. This means that a comprehensive approach is needed to deliver propagation models that would facilitate the optimum placement of RSUs.
4.2 Backhaul Issues
Another key issue is backhauling data to the central servers or the Cloud Infrastructure. The testbed uses two backhauling mechanisms: the RSUs on the Hendon Campus are connected to the University’s 1 Gbps Ethernet network, while the RSUs on the motorway use LTE to backhaul the data. Preliminary tests using pings show that the latency from the RSUs on the Hendon Campus are in the order of fractions of milliseconds while the latency of the RSUs on the A41 are in the order of tens of milliseconds. This means that LTE is too slow to run life-critical or remote control applications. The other alternative is using fibre, but fibre is expensive and ensuring that it is accessible throughout the road infrastructure is challenging.
4.3 The use of 5G in Vehicular Networks
5G, the next generation of mobile phone technology, promises to have low latency and so can be used in vehicular environments. In order to investigate this further, it has been decided to build a new Connected Vehicle testbed, along the Strand in Central London at King’s College London (KCL). This effort, known as the Central London Testbed Project (CLTP) will examine the interaction between the emerging 5G standards and its support for vehicular environments. CLTP will be connected to the MDX Testbed using Software Defined Networking (SDN) and the Joint University Academic Network (JANET) to form a Federated Cloud System as shown in Figure 6. This system is expected to be fully operational by August 2017.
The Central London VANET/5G Cloud Testbed is being built to support Connected Services in which a large number of diverse applications between mobile users and the Internet will be tested including high quality video steaming, large file downloads via the Web and a high-quality network music server. In addition, directly interactive applications, such as video-conferencing, which will allow mobile users in vehicles to talk to each other via the Federated Clouds will also be instrumented. The aim here is to come up with values of key parameters such as bandwidth, latency, jitter and profiles of errors tolerance that can be used to quantify the performance of these applications in a connected vehicular environment and thus to give 5G developers real figures at which they should aim as they develop the new communication standards.
4.4 Building an Application Framework for the Connected Vehicle Space
In order to produce an environment to develop applications for Connected Vehicles using the VANET/5G testbeds, a new application framework known as the Connected Vehicles Application Framework (CVAF) is being developed. The architecture is shown in Figure 7.
These testbeds will gather information from vehicles, including speed, location and direction of travel as well as braking pressure, wheel acceleration, fuel usage and readings from the Electronic Control Unit (ECU) of the vehicle.
The testbeds will also gather information from the Roadside Units as well as Mobile Access Points and Base Stations of mobile operators. Geo-spatial information, including the location of buildings, roads, etc., in a given area will also be looked at. All this information from the Testbeds will be stored on the VANET Cloud Servers.
Intelligent Information Platform
Using the Apache Hadoop Framework (AHF), and the data stored on the VANET Cloud Servers, techniques will be developed to look at data in various ways including:
Location: There will be the ability to look at all the traffic at a given location or in a specific region over any period of time.
Vehicles: The system will also allow us to look at a particular vehicle or group of vehicles.
Users: The system will allow us to look at the movement of a user or a group of users.
Infrastructure: The system will allow us to examine readings from a given RSU or Mobile Access Point or Base Station. This will allow us to look at better deployment strategies for the transport and communication infrastructures.
We intend to work on the key applications shown in Figure 7. Firstly we will work on delivering connected services to the car, followed by vehicular control technology using Virtual Reality (VR) and Haptic Techniques.
In conclusion, we would like to thank Transport for London (TfL) and Mobius Network for their support for this project. All technical enquires concerning this article should be sent to Dr Glenford Mapp at firstname.lastname@example.org.