TU BRAUNSCHWEIG
| Carl-Friedrich-Gauß-Fakultät | Informatik
Informatikzentrum

Demonstrator for Consensus Protocols in Wireless Ad-Hoc Networks

BetreuerMartin Wegner
Wenbo Xu
ProfessorProf. Dr. Rüdiger Kapitza
ZweitgutachterProf. Dr.-Ing. Lars Wolf
Projektccc
IBR GruppeCM (Prof. Wolf), DS (Prof. Kapitza)
ArtTeamprojektarbeit
Statusabgeschlossen
Beginn2016-08-01

Description

Consensus protocols originally stemmed from data center scenarios where fixed networks with high bandwdiths and low delays are prevalent. They have been used in these environments to achieve Byzantine Fault Tolerance (BFT), e. g., for state machine replication.

Within the project CCC we seek to apply consensus protocols to the domain of Wireless Networks, especially Ad-Hoc networks such as Vehicular Ad-Hoc Networks (VANETs). These networks come with their own set of challenges: due to the load or loss on the wireless channel, delays have comparably higher upper bounds — messages will get lost in some cases, and retransmissions occur on a regular basis.
Consequently, a deployed consensus protocol has to deal with all these cases, i. e., it needs to be able to handle possibly duplicate messages, longer delays, etc. Moreover, they also may be attacked more easily since an attacker simply might inject or replay messages of the protocol on the wireless channel.

Within this thesis, you shall setup a first hardware demonstrator for a consensus protocol implementation in the wireless domain: A set of provided hardware nodes (e. g., Raspberry Pis) have to be setup with an existing, given Java implementation of a consensus protocol. In addition, each node will be equipped by a hardware sensor, e. g., for light. All nodes communicate in an Ad-Hoc network set up using IEEE 802.11 b/g/n.
The consensus protocol shall then be used to reach consensus among these nodes on the current sensor value.

To prove the correctness of the demonstrator and also enable future tests of the robustness of the protocol with respect to the wireless environment, a set of test cases shall be set up and evaluated:

  • show how the consensus protocol reacts to corrupted/false sensor data, i. e., in the case a light sensor is covered by an obstacle, etc.,
  • evaluate misbehaving nodes, such as a node dropping all messages (black-hole attacker), or a node manipulating his own sensor readings or those of other nodes when running the protocol,
  • devise a setup where the path loss rate and consequently also the retransmissions increase rapidly, e. g., by positioning appropriate obstacles in the line-of-sight between two nodes or similar.

Requirements

For this thesis, extended knowledge of the Java programming language as well as the Linux Operating System is paramount. Prior work with embedded systems (Raspberry Pis, GPIO, hardware sensors) as well as experience with C/C++ certainly is beneficial.

If you are interested in this thesis, please contact Martin Wegner or Wenbo Xu via mail, including the following information:

  • Current major (i. e., computer science, information systems technology, etc.)
  • relevant passed courses
  • semester
  • prior knowledge (see above)

Links


aktualisiert am 17.05.2017, 12:59 von Martin Wegner
printemailtop