Abstract:
Vehicular Edge Computing (VEC) is an emerging optimistic computing paradigm that
aims to decrease back-haul network traffic and service latency by extending cloud
service provisioning to network edge vehicles. As the Internet of Things (IoT) and
telecommunication industry continue to progress, several innovative applications/fea tures including image-aided navigation, autonomous vehicles, and face recognition are
starting to appear. Cooperative, linked, and autonomous vehicles present in the net work generate various types of resource-intensive and delay-sensitive tasks. These au tomotive applications demand a minimum processing delay along with a large amount
of computational power. However, these power and resource-constrained vehicles might
not be able to process these vehicular applications. VEC is considered an innovative
paradigm for enhancing vehicles’ performance by executing various applications to the
edge cloud. VEC benefits users by providing computational resources to their proximity
thus decreasing the latency, but VEC has limited computational capacity and cannot
process all tasks for a larger number of requests. Moreover, each task has individual
requirements such as some tasks are delay-sensitive, some tasks require lower energy
consumption, some tasks require real-time information, and some vehicles might have
low battery levels. Also, in a highly dynamic topology, smaller task execution across
a VEC server results in higher energy consumption and communication latency. It
is required to optimally select the tasks for offloading to address these issues. This
research proposes efficient task-centric and classification-based resource allocation and
offloading strategies based on individual users’ requirements and task prioritization.
In this thesis, we have discussed two schemes. The first scheme aims to lower the sys tem cost in terms of processing time while considering the tasks’ priority. The second
scheme suggests an algorithm to lower the value of both processing delay and energy
consumption while considering tasks’ priority, energy requirements, and battery level
of vehicles. These proposed schemes offload the tasks to the VEC server according to
the individual task requirements along with reducing system costs.
