Placement of Distributed Services on Vehicle Clusters for Sensing Applications

Sharma, Kanika (2023) Placement of Distributed Services on Vehicle Clusters for Sensing Applications. Doctoral thesis, SETU Waterford.

[thumbnail of phd_thesis_KanikaSSharma_SETU_31stJan.pdf] Text

Download (18MB)


Vehicular fog computing (VFC) is an extension of cloud and mobile edge computing aimed to utilise the rich sensing and processing resources available in vehicles. The emergence of VFC was motivated by the need to reduce latency in delay-sensitive applications, which makes it infeasible to deploy applications on the cloud. Thus, there emerged a need to deploy data processing services close to the source of data generation to reduce response time and bandwidth usage in uploading collected data to the cloud. Most modern vehicles in the near future will be equipped with plentiful sensing and processing capacity to make sophisticated decisions related to autonomous driving. The aim of this study is to utilise the under-utilised resources on these vehicles to deploy data-intensive, over-the-top services on a group of closely-moving vehicles. With the emergence of data-driven applications developed in the process of urban informatization, there is a need to collect and process data in a resource-efficient manner. In this work, we consider applications where vehicles gather and process data for surveillance purposes such as studying the interaction of users with roadside cafes and gas stations etc. or detecting vulnerable pedestrians to increase commuter safety. Our work introduces methods and techniques to use the historic mobility pattern of vehicles to address the challenge of dynamism and instability in the vehicular network. To determine the availability of these resources, a stochastic mobility model is utilised, to select nodes with similar mobility patterns. Then a distributed and flexible service model and a mobility-aware infrastructure model are designed that are compatible with an unstable, non-static network. These distributed services are scaled in real-time and placed as multiple instances on the selected vehicular cluster to make the services more robust. The service scaling and placement problem is modelled as a bi-objective, constrained optimisation problem with the objective of efficient resource utilisation. To place the service chains efficiently on the vehicular clusters, a community detection-based cluster selection scheme and graph-based service placement heuristics are introduced. The feasibility of the study is presented by demonstrating results from extensive simulations using different resource and mobility profiles of vehicle clusters. The results of the performance of our service scaling and placement scheme indicate that it performs better than competing schemes in terms of resource utilisation.

Item Type: Thesis (Doctoral)
Uncontrolled Keywords: Distributed services, Vehicle clusters, Applications
Departments or Groups: *NONE OF THESE*
Divisions: School of Science > Department of Computing, Maths and Physics
Depositing User: Derek Langford
Date Deposited: 29 Mar 2023 13:51
Last Modified: 29 Mar 2023 13:51

Actions (login required)

View Item View Item