Channel-error and collision aware, secure multihop ad hoc wireless networks

  • Authors:
  • Jun Cheol Park

  • Affiliations:
  • The University of Utah

  • Venue:
  • Channel-error and collision aware, secure multihop ad hoc wireless networks
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

This dissertation develops efficient and secure multihop ad hoc wireless networks. A good multihop ad hoc wireless networks design must address the challenges posed by coexisting channel-errors due to error-prone wireless channels and complex collision situation due to transmission interference. Specially, the interaction of collisions and channel-errors makes the assessment of each quantity very difficult, resulting in poor performance of important components of multihop ad hoc wireless networks. These components include ad hoc routing and multirate adaptation. In this dissertation, we develop two new ad hoc routing protocols and two novel multirate adaptation schemes to improve the performance of multihop ad hoc wireless networks by carefully considering both channel-error and collisions. We develop an ad hoc routing algorithm, Minimum Distance-2 (MIND2), in the context of a hybrid wireless architecture that uses 802.11-based ad hoc wireless networks in conjunction with the cellular multicast networks. Our ad hoc routing algorithm uses minimum distance-2 vertex coloring to model the effect of transmission interference in determining achievable throughput of ad hoc paths. Using ns-2 simulation of 3G HDR BCMCS networks, we show that our algorithm increases the multicast receiver goodput by up to 280% in comparison to BCMCS that does not use ad hoc paths. We also design a new path metric, Expected Path Throughput (EPT) for accurately finding high-throughput paths in multihop ad hoc wireless networks. Our metric is based upon a realistic and practical model that considers both transmission interference and channel-error by (i) determining transmission contention degree of each link as a function of wireless channel-error, (ii) quantifying the impact of the wireless channel-error on medium access backoff, and (iii) considering possible concurrent transmissions when two links do not interfere with each other. We find that EPT can accurately determine the achievable data rates of ad hoc paths, thereby significantly outperforming the other existing metrics. This dissertation also presents two different novel multirate adaptation schemes that efficiently differentiate between collision loss and channel-error loss. First, we design a Cross Layer Multirate Adaptation (CROMA) scheme to improve the performance of multihop wireless networks. Our scheme effectively distinguishes loss due to collisions from those due to channel-errors by using a cooperative exchange of loss information between senders and receivers in the wireless network, based on the phenomenon of physical capture. Second, we develop another novel multirate adaptation scheme, Reduced Packet Probing (RPP), that enables a sender node to effectively approximate channel-error loss in the presence of collisions without relying on any special information from the physical layer at receiver nodes. We evaluate our RPP multirate adaptation scheme in the wireless Emulab testbed and show that it achieves up to 53% higher TCP performance in comparison to the existing algorithms. In addition to performance, multihop ad hoc networks face serious security challenges. Due to lack of an infrastructure and a well-defined perimeter, ad hoc networks are susceptible to a variety of attacks. We develop a scheme that secures multihop ad hoc wireless networks against data injection attacks by placing firewall functions at strategic locations. We also develop an architecture to detect attacks at victim nodes and determine the locations of attackers. In our approach, victim nodes can themselves determine that they are being attacked when they receive unwanted packets. Our architecture uses a separate control network (a cellular network in our research) in conjunction with an ad hoc network, to provide a provable attack detection mechanism.