Abstract
Efficiency Study of Sybil Attack on P2P Botnets
Yuhang Luo
The main objective of this thesis is to modeling and analysis of Kademlia based
Botnets in order to study the efficiency of Sybil attack on such botnets.
We start by researching the structure of Kademlia and specially its look-up procedure,
i.e. the process how a node find a desired target node in the Botnet. For the simplicity
of analysis, two assumptions are made: a) node ID space is full filled; b) a Sybil node
replies a fake triple when it is queried.With these assumptions, the probability jumping
functions and jumping matrices are derived. By adding the distribution of Sybil
nodes, we obtain the probability that target nodes are found successfully(Psuccess).
We then show numerical results of the distribution of Psuccess with different system
parameters. From the results, we can obtain some insight on how the parameters
affect the efficiency of Sybil attack. Among all these parameters, we find that �,
which is known as the number of nodes the initial node requries, is the key parameter
of Kademlia based botnet. We also discuss how the triples a node keeps for every
distance (k) and the total number of nodes in botnet (n) will affect Psuccess.
Based on our model and numerical results, we will draw some conclusions on how to
make P2P botnet more robust or more vulnerable in facing Sybil attacking.