| Advancements in wireless technologies will allow users to communicate at any time at anyplace. Not only will wireless networks becomes more available to users, but adversaries will be able to easily obtain wireless devices. Unfortunately, the open nature of the wireless medium and the lack of a well-defined security architecture make wireless networks particularly vulnerable to attacks. In particular, the adversary's ability to inject, modify and drop traffic in a wireless network can facilitate a variety of attacks, and can serve as the basis for conducting denial of service (DoS) attacks.; In this dissertation, we have studied methods for coping with several forms of attacks in wireless networks that involve the injection of traffic, and which can lead to denial of service attacks. Specifically, we have examined issues related to: protecting multicast traffic by developing authentication technologies that are resistant to flooding-style DoS attacks; detecting device spoofing in wireless networks and the corresponding anomalous traffic that ensues; and the disruption of routing in wireless ad hoc networks. Our solution to multicast flooding DoS includes an improved authentication scheme, which we call Staggered TESLA, that still employs the delayed key disclosure principle, but achieves resilience to DoS and provides multi-level security for multicast applications. In order to address spoofing, we have introduced a framework for detecting anomalous traffic that is based upon forge-resistant relationships. Investigations have shown that it is possible to create these relationships by introducing fields into packets, or by using the intrinsic properties associated with the transmission and reception of packets. These relationships can be further used to construct classifiers that provide multi-level threat assessment. We have addressed routing disruption attacks in wireless ad hoc networks by introducing a new secure routing protocol that efficiently utilizes authentication fields in routing messages. In particular, the protocol we have developed, which we call SEAR (the Secure Efficient Ad hoc on-demand Routing protocol), primarily uses symmetric cryptography and requires asymmetric cryptography only for the initial bootstrap. Compared to existing secure AODV routing protocols, which utilize expensive public key cryptography, SEAR provides better security with significantly less overheads. |
