Experimental comparisons between SAODV and AODV routing protocols

Yuxia Lin, Amir Hamed Mohsenian Rad, Vincent W. S. Wong, Joo-Han SongWmunep 2005 · DOI

TL;DR

This paper presents a real-world experimental comparison of the AODV and SAODV routing protocols in mobile ad hoc networks (MANETs). They found that SAODV effectively prevents message tampering and data dropping attacks, even with the overhead introduced by security mechanisms, and also characterized the "communication gray zone" in wireless links.

What Problem Does This Paper Address?

Mobile Ad hoc NETworks (MANETs) are decentralized networks where devices communicate directly without fixed infrastructure. Routing protocols are essential for devices to discover and maintain paths to each other. While protocols like AODV are well-established, they are vulnerable to security threats like message tampering and data dropping, especially in dynamic, open environments. Existing security analyses often rely on simulations, leaving a gap in understanding how these protocols perform in real-world testbeds with actual mobility and wireless impairments. This paper aims to bridge that gap by experimentally comparing a standard protocol (AODV) with a secure variant (SAODV) in various realistic scenarios.

Methods

The authors implemented SAODV by modifying an existing AODV implementation (AODV-UU) for the Linux environment. SAODV uses digital signatures to authenticate non-mutable fields in routing messages and hash chains to authenticate the hop-count field, thereby protecting against tampering and replay attacks. They also implemented an attacker module to simulate message tampering and data dropping.

The experimental setup involved three phases:

  1. Simulation: Using the ns-2 network simulator.
  2. Indoor Emulation: Using a custom mobility emulator (MacSim) that leverages Linux's iptables to simulate link breakages based on pre-generated mobility traces from ns-2. This allows for controlled, repeatable indoor experiments.
  3. Outdoor Experiments: Conducting tests on a small-scale, real-world testbed consisting of 10 laptops in a 250m x 100m rugby field.

They evaluated both UDP and TCP traffic under these conditions, both with and without the simulated attacker.

Datasets

Not discussed in the paper. The authors generated their own mobility traces using ns-2 for simulation and indoor emulation, and conducted real-world experiments with 10 laptops.

Key Insights and Ideas

  • Real-world security matters: SAODV's effectiveness in preventing attacks was validated not just in simulations but also in a physical testbed, demonstrating its practical utility.
  • Security overhead is manageable: Despite the added overhead of digital signatures and hash chains, SAODV maintained competitive performance, particularly in its ability to deliver packets when under attack.
  • The "communication gray zone" is a real phenomenon: The experiments quantified the existence of zones where nodes might receive some packets but not others, impacting routing reliability. This zone's size is dependent on transmission rates.
  • Indoor emulation is a useful proxy: The indoor emulation results closely matched simulation findings, suggesting it's a viable method for testing MANET protocols when full-scale outdoor tests are impractical.
  • TCP resilience: TCP traffic showed some resilience to attacks compared to UDP, as TCP could potentially recover lost packets or reroute traffic more effectively over time, especially in a dynamic environment.

Key Results

  • Attack Prevention: SAODV significantly improved packet delivery ratios compared to AODV when an attacker was present, effectively preventing message tampering and data dropping attacks in all tested environments (simulation, indoor, outdoor) for both UDP and TCP traffic.
  • Routing Overhead: SAODV introduced additional overhead (in bytes) for routing packets (RREQ, RREP, RERR) due to security fields (signatures, certificates, hashes). However, this overhead did not drastically degrade performance, especially for TCP.
  • Communication Gray Zone: The measured "gray zone" width was approximately 9m at 2 Mb/s and 18m at 11 Mb/s. The transmission range (where packet delivery is >90%) was around 110m at 2 Mb/s and 100m at 11 Mb/s.
  • UDP vs. TCP: UDP traffic was more severely impacted by attacks, with packet delivery ratios dropping significantly. TCP throughput also decreased under attack but showed some recovery.

Limitations and Future Work

  • Small-scale testbed: The experiments used only 10 laptops, which limits the path lengths and the number of alternative paths that can be explored. The authors plan to incrementally increase the number of nodes.
  • Single attacker: The experiments primarily focused on a single attacker. Future work will investigate the impact of multiple attackers.
  • Indoor emulation realism: The indoor emulation used simplified ON/OFF link states and mobility traces derived from simulations. Future work aims to use actual outdoor mobility traces for more realistic indoor emulation.
  • Driver limitations: Issues with wireless card drivers (especially for ad-hoc mode and signal strength reporting) limited some advanced testing, like using signal strength for gray zone mitigation.
  • Future research: The authors plan to compare SAODV with other secure routing protocols, increase testbed scale, and explore cross-layer optimization techniques.

Summary updated

BibTeX
@inproceedings{wong2005experimental,
  author = {Vincent Wong and Yuxia Lin and Amir Hamed Mohsenian Rad and Vincent W. S. Wong and Joo-Han Song},
  title = {{Experimental comparisons between SAODV and AODV routing protocols}},
  booktitle = {Wmunep},
  year = {2005},
  doi = {10.1145/1089737.1089757},
}
Figure