Vehicular Communication for High Precision Cooperative Localisation
Sujet proposé par
Directeur de thèse:
Unité de recherche
Laboratoire de recherche d'EURECOM
Domaine: Sciences et technologies de l'information et de la communication
The automotive industry completed in 2015 the specification of vehicular communications standard for the first generation of Cooperative Connected Vehicles (C2V) applications. It is currently moving toward the second generation of C2V applications, including Autonomous Driving, or Safety of Vulnerable Road Users. These applications will require high precision positioning services at the sub-meter level, which is not yet available to C2V applications by any mass market GNSS technology, such as GPS, GLONASS, Galileo etc, operating in any environment.
This opens new horizons to benefit from C2V advanced communication and processing capabilities to develop distributed rather than centralized, relative rather than absolute, positioning systems. Yet, considering that reliable data exchanges between C2V have been shown to be already challenging for small individual data volumes, the significantly higher expected data volume (contextual data rather than just GPS positions), and the need for stringent latency and dependability levels will require a new way of thinking and designing C2V data communication for such positioning systems.
The goal of this position is to propose, model and analyse new dependable C2V communication strategies, satisfying jointly the high volumes of data required to be exchanged between vehicles, as well as new C2V communication constraints required by cooperative positioning services for highly autonomous vehicles.
The work will be carried out in the framework of the European project HIGHTS . The HIGHTS project aims at providing sub-meter (<0.5m) cooperative positioning for autonomous vehicles, and its consortium comprises strong industrial partners as well as a number of world-class universities. It is funded by the European Commission under the H2020 Framework Program. M. Irfan Khan will actively participate to the progress of the project.
In C2V, cooperative communication represents the cooperation between vehicles through the exchange of local and/or perceived data (position, kinetics) in a fully ad-hoc manner. Unlike infrastructure or centrally controlled communication, in C2V scenario there is no infrastructure or central node that has a global view of the network to optimally schedule each vehicle’s communication. In the absence of control, cooperative communication typically reaches channel saturation and leads to a drastic drop of the communication reliability. Controlling the communication parameters (sometimes known as ‘scheduling radio resources’ in WLAN and 4G) is known as ‘Decentralized Congestion Control (DCC) in C2V, and represents a research challenge, which is at least as complex as that of optimal scheduling and resource allocation in WLAN/cellular networks.
Specifically, the focus of this position will be to come up with smart and adaptive DCC mechanisms in order to guarantee the communication requirements for Cooperative Positioning (CP) of vehicles, for use cases such as Highly Autonomous Driving (HAD) and/or detecting Vulnerable Road Users (VRU). The existing congestion control mechanisms (at state their current state) have been developed for DAY ONE Intelligent Transport System (ITS) applications, e.g. Road Hazard Warning,Intersection-collision avoidance, and Cooperative awareness. For these applications, positioning was assumed to be ‘perfect’ (ignoring the GNSS errors). For DAY TWO applications, such assumptions may not be practical and not only GNSS positioning errors must be included, but advanced positioning must be proposed to reach a precision beyond GNSS. These applications will require stringent awareness information from cooperative vehicles, and on the other hand cooperative vehicles may also be seen as an opportunity for advanced positioning systems, through their ability to exchange and share data.
One challenge is that CP will add additional communication requirements, which from initial analysis has been shown to be contrary to those of HAD and VRU. A DCC mechanism will therefore have the responsibility to not only guarantee dependable vehicular communication to transmit the positioning data required by HAD and VRU, but also guarantee that such information is highly precise. This aspect will be considered as the first research problem to be addressed by M. Irfan Khan in his doctoral work.
A second challenge is that the current C2V standards have been developed considering only one single class of traffic (i.e. periodic awareness message, a.k.a Cooperative Awareness Message (CAM) in EU and Basic Safety Message (BSM) in US). It is highly expected that HAD, VRU as well as CP will require at least some extensions to the currently defined messages, if not entirely new messages and classes of traffic. In particular, DCC mechanisms are currently not adapted to meet the heterogeneous objectives from different traffic classes. Prioritization will not only be required from the MAC layer, but also from the communication parameters (coding rate, range, rate. etc). Such challenges have already been identified in literature work, but need to be evaluated from a multi-disciplinary perspective involving CP and HAD/VRU. This will be considered as the second research objective to be investigated during M. Irfan Khan’s doctoral work.
A third challenge is that current C2V applications only consider vehicle-to-vehicle or vehicle-to-infrastructure communication. HAD and VRU will also include different classes of communicating ‘nodes’, such as pedestrians, motorcycles or other smart wearable devices. Considering the particularities of these new players, it is also highly likely that they will require different classes of traffic. But most challengingly, they will generate additional traffic load on the shared vehicular channel, which through existing research has been shown to be already quite saturated due to the communication among vehicles only. Therefore, DCC mechanisms will have the responsibility to provide fair access to the vehicular channel, and similarly additional channels or technologies might need to be used to sustain the thirst of data that future HAD/VRU applications will require. Considering ITS-G5 technologies, two other channels are available, each of them with constraints that need to be considered, but with a major advantage that no existing specifications on MAC or DCC use those channels. This makes these channels a promising opportunity for clean slate design adapted to the requirements of HAD/VRU and CP. Similarly the possibility to use PHY and MAC layer technologies with higher throughput and control (e.g. 802.11ax) will be analysed. This will be considered as a third problematic to be investigated by M. Irfan Khan during his doctoral work.
It is also highly likely that other challenges will be discovered by M. Irfan Khan during his doctoral work, in particular during and after the first phase dedicated to related work and state-of-the-art (SotA) analysis. Whenever they arise, these challenges will most certainly require new communication strategies within the current system (e.g. transmit power, rate, channel), but also from external systems (ex. cellular/PAN system, IoT/M2M systems, GNSS), which M. Irfan Khan will integrate in his work if identified as correlated.
In a first phase, M. Irfan Khan will therefore specify his research problem by conducting a thorough SotA investigation in technologies and standards, in particular focusing on a gap analysis between the specification of current technologies (ITS-G5, 3GPP, PAN) and the requirement from future HAD/VRU and CP. He will also extend his SotA to other systems (IoT, GNSS) and players (vehicles, motorcycles, pedestrians) in order to obtain a holistic view of the challenges of providing dependable cooperative communication and positioning for HAD/VRU. M. Irfan Khan will also investigate the current SotA of DCC mechanisms developed in EU, US and Asia.
In the second phase, M. Irfan Khan will structure his doctoral work in a three step methodology:
(i) Abstraction, modelling and analysis of connectivity properties of cooperative vehicular networks
(ii) Methodology for C2V communication and congestion control mechanisms adapted for cooperative positioning, VRU and HAD
(iii) Extensions of C2V communication protocols for cooperative positioning services
For the first step, M. Irfan Khan will abstract the basic communication requirements from the HAD, VRU and CP (in terms of reliability, range, delay, message size), and then analyse the fundamental behaviour of the state-of-the-art DCC mechanisms (such as the CAR-2-CAR, LIMERIC or PULSAR), to estimate their conformance to the requirements. M. Irfan Khan will follow a cyber-physical system (CPS) modelling, and define the role of the physical blocks (e.g. channel load) and cyber block (e.g. actuating communication parameters). Such methodology will allow to identify the parameters leading to system instability, which would be dramatic in case of HAD and VRU.
In the second step, M. Irfan Khan will apply the outcome of the previous phase to propose an actuating methodology that would adjust transmit power or rate considering the different classes of traffic. One approach will be to consider randomized strategies satisfying the joint requirements. Another approach will be to explicitly indicate what kind of data will be required by which cooperative vehicles following which kind of traffic class in order to avoid overloading the channel. The objective will be to define innovative communication strategies (CPS sensing – actuating) that could firstly satisfy the requirements of HAD/VRU and CP, and secondly would provide stable behaviour of the system as a whole, and finally would be supported by the vehicular communication channels.
In the last step, M. Irfan Khan will move to the implementation phase, where he will apply the extensions to current protocols and technologies. One key aspect in these phase will be to define one or multiple evaluation scenarios spanning different environments, but also different communication and positioning conditions. Another key aspect will be to define the testing methodologies in order to obtain reproducible, and statistically meaningful results. Then, the evaluation will be conducted on simulators adapted to C2V (iTETRIS , ns-3 ), and also implemented on HIGHTS C2V experimental testbed, including test vehicles.
Whenever applicable, M. Irfan Khan’s work will be brought to the standardization bodies, as well as integrated in related work conducted by HIGHTS partners, in order for M. Irfan Khan’s work to benefit from outcome from others, but also let his work impact those of others.