Please use this identifier to cite or link to this item:
Title: Deep Learning-based Data Dissemination Scheme in Content-Centric Internet of Vehicles
Authors: Gulati, Amuleen
Supervisor: Kumar, Neeraj
Keywords: Content-centric Networking, Convolution Neural Network, Deep Learning, Internet of Vehicles
Issue Date: 14-Aug-2018
Abstract: The IP-based IoV networks aimed to provide vehicle-to-vehicle communication to enable information exchange between vehicular users. However, due to the address-content binding mechanisms used in IP-based protocols, the IP-based networking proves to be costly and inefficient in terms of overhead incurred due to DNS lookups, load balancing, etc. The Content Centric Networking (CCN) technology provides an alternative by enabling the users to share the information based upon the content, thereby providing a content-centric model as opposed to hostcentric model of IP-based networks. Previous studies have shown that CCN uses transfer of named data and in-network caching mechanisms to enable communication in an infrastructureless environment. But its broadcast-based access mechanism has given rise to new problems related to network congestion and increased delays. This problem is more severe in vehicular network scenario where the network nodes are mobile and require dynamic addressing. This thesis addresses this issue by putting forward the design of a data dissemination scheme in an IoV scenario using Content-Centric networking architecture is proposed. The proposed scheme works in three phases: (1) Firstly, the vehicles will be screened based upon the remaining energy of the vehicles. This is done to ensure that the vehicles do not run out of energy during data transmission. (2) In the next step, the connection probability of each potential vehicle pair is computed with an aim to promote only stable and long lasting connections to the next level. (3) In the last step, a convolutional neural network (CNN)-model is used for community detection by using the social layer information of vehicular users. CNN is used to identify the ideal vehicle pairs, which can share data to ensure minimum delay and high data availability. This application is based on the assumption that the users which belong to the same community are more likely to download the same type of information and hence, rather than using cellular links for data transfer, the vehicles can get the data from other nearby vehicles belonging to the same community and hence minimize the cost of cellular link usage and delay. To improve the performance of CCN, several hyper parameters are adjusted and tuned to achieve optimal performance
Appears in Collections:Masters Theses@CSED

Files in This Item:
File Description SizeFormat 
final_thesis_amuleen gulati.pdf3.94 MBAdobe PDFThumbnail

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.