Thesis

Optimizing Joint Radio and Computing Resource Allocation in Vehicular Edge Computing

V2X (Vehicle to Everything) is a key component of today's Intelligent Transportation Systems (ITSs). It controls how vehicles coordinate with one another and communicate with other objects that might have an impact on the environment. These objects include other vehicles, pedestrians, and surrounding infrastructure such as traffic lights. ITS applications are feasible thanks to the exchange of some predefined data messages via wireless communications and processing them by a small computing node embedded in the vehicles or by a remote robust server in a distant cloud when a large computing capacity is required. However, certain critical services require relatively large computing and storage resources than what a vehicle may have. Besides, some of these services are sensitive to the delay, and forwarding the pieces of information to a remote central cloud for processing may be inappropriate. To address this issue, the multi-access edge computing (MEC) for 3GPP 5G Cellular vehicles to everything (C-V2X) has been proposed [1][2]. MEC is a flexible cloud environment based on the concepts of Network Function Virtualization (NFV) and Software Defined Networks (SDN); it is located at the edge of the 5G network, in proximity of the end users. By acting at the edge of the network and coupled with the service provider’s network infrastructure, MEC enables data processing to be done faster than the backend system hence permits to provide services with low latency and high bandwidth. The utilization of MEC in vehicular networks is known as Vehicular Edge Computing (VEC) and is considered an important research direction that has recently received a lot of attention from the scientific community [3] [4]. However, as studied in [5-8] the radio resource allocation in VEC is typically associated to the computational resource allocation. One of the key objectives of joint resource allocation for vehicular edge computing is the decrease of overall latency, which includes the processing time at the edge server and the transmission time at the radio interface. To achieve this goal, the allocation of radio resources, which affects the transmission time, must be managed efficiently. In addition, allocating high computational resources is the best choice to reduce the processing time at the edge server. The goal of this PhD in to deeply investigate this research topic and to propose efficient radio and computing resource allocation using the concept of VEC to reconfigure radio and computing resources rapidly and dynamically in response to vehicles mobility. Furthermore, we aim in this PhD at building realistic and scalable simulation scenarios with heterogenous service types having different QoS requirements (e.g., bandwidth and delay). Simulation can be done using the Simu5G open-source project [9] [10] as a base simulator for the V2V/V2I radio communication. This simulator incorporates 4G and 5G New Radio access based on the OMNeT++ framework [11]. It is written in C++ and is fully customizable with a simple pluggable interface. Simu5G integrates also a MEC module. one can also develop new modules that incorporate advanced algorithms and protocols, leveraging data science techniques. Specifically, simulations can generate synthetic data, which can be utilized to train machine learning algorithms for predicting resource allocation and configuring proactive SDN-based flow rules. Such an approach can enhance the efficiency and effectiveness of resource allocation in vehicular edge computing. Another interesting aspect to consider is the comparison between machine learning-based techniques and traditional optimization approaches. This comparison can shed light on the strengths and weaknesses of each approach and provide insights into which one is more suitable for specific use cases in vehicular edge computing.