Files in this item

FilesDescriptionFormat

application/pdf

application/pdfJonathan_Ponniah.pdf (1MB)
(no description provided)PDF

Description

Title:A clean slate approach to secure wireless networking
Author(s):Ponniah, Jonathan
Director of Research:Kumar, P.R.
Doctoral Committee Chair(s):Kumar, P.R.
Doctoral Committee Member(s):Hu, Yih-Chun; Vaidya, Nitin H.; Srikant, Rayadurgam
Department / Program:Electrical & Computer Eng
Discipline:Electrical & Computer Engr
Degree Granting Institution:University of Illinois at Urbana-Champaign
Degree:Ph.D.
Genre:Dissertation
Subject(s):Multi-hop wireless networks
Utility maximization
Game theory
Security
Secure clock synchronization
Abstract:Traditionally, wireless network protocols have been developed for performance. Subsequently, as attacks are identified, patches or defenses have been developed. This has led to an ``arms race,'' where one is never confident about what other vulnerabilities may be exposed in the future. We seek to reverse this process. We identify a set of axioms describing a model, under which we develop a secure utility optimized network. Our results rest on the axioms, and can be attacked only to the extent that the axioms can be challenged. We present a complete suite of protocols, taking a wireless network all the way from startup to optimality. These protocols are not just individually secure; they are holistically secure, that is, there are no gaps between them that can be attacked. The approach considers a group of wireless nodes some of which are ``good,'' and the rest, ``bad.'' The good nodes seek to form a functioning wireless network, operating at a high level of utility. The bad nodes know the identities of the good nodes but not conversely. Moreover, unlike their good counterparts, the bad nodes are capable of full centralized cooperation and collusion. On the other hand, the good nodes arrive on the scene unsynchronized, uncoordinated and ignorant of the others' intentions. We introduce a distributed protocol suite that enables the good nodes to proceed all the way from birth to a min-max utility optimal network, where the minimization is over all bad behaviors of the bad nodes, and the maximization is over all protocols followed by the good nodes. That is, the good nodes form a functioning, reliable network from startup, in the face of any sustained cooperative attack mounted by the bad nodes. We show that the protocol overhead occupies an arbitrarily small fraction of the total operating lifetime. We prove that our protocol realizes a nearly optimal level of utility. Our protocol supersedes a considerable amount of previous work that deals with several classes of attacks such as the following: man-in-the-middle, wormholes, dropping packets, Byzantine behaviors, disruption of timing events, presenting false topologies, etc. More importantly, this protocol suite obviates the need to identify all of the other types attacks that can potentially be carried out by colluding malicious nodes, for there are many. Instead, under this protocol, the malicious nodes cannot reduce the utility of the network any further than they could by either just jamming and/or cooperating with the protocol. At a broader level, our approach presents a model-based approach to secure protocol development, as an alternative to an arms race type of approach.
Issue Date:2014-01-16
URI:http://hdl.handle.net/2142/46565
Rights Information:Copyright 2013 Jonathan Ponniah
Date Available in IDEALS:2014-01-16
Date Deposited:2013-12


This item appears in the following Collection(s)

Item Statistics