Resource-Aware Epidemics in P2P Systems

March 16, 2012 – ,

Overview

Epidemic protocols, such as gossiping or flooding, represent an appropriate class of communication mechanisms to collect and distribute information in decentralized systems. Especially in large-scale networks, such as P2P systems, these protocols are deployed, since they exhibit high robustness and stability and are insusceptible against the inherent dynamic of such networks. Therefore, many communication mechanisms, which collect and distribute any kind of information, rely on this class with its diverse protocol implementations.

Problem Statement

Robustness and stability of these protocols come at the price of increased costs. While a node can control and choose to how many neighbours it may send its information, it cannot control for how many messages from its neighbours it might be the receiver. The increased and uncontrollable message overhead might not be critical for the transmission of small data or in homogenous networks, but it might become a problem for large data or especially in heterogeneous networks, where we have different types of devices with scarce resources (e.g. battery power on smart phones).
Therefore, an epidemic protocol should be aware of the different devices and take the resources of a device into consideration when transmitting data.

Tasks

  • Get into the topic of epidemic protocols, especially gossip-based protocols
  • Investigate how to integrate a certain kind of resource-awareness into these protocols
  • Design an own resource-aware protocol
  • Evaluate the protocol on top of a simulated heterogeneous P2P system

Prerequisites

  • Motivation and analytical skills to investigate on interesting research topics
  • Basic knowledge about computer networks (e.g. from KN x or TK x)
  • Basic knowledge about P2P systems
  • Good programming skills in Java

download corresponding tendering

Keywords: P2P;Gossiping;Flooding;Heterogeneity

Research Area(s): Self-organizing Systems & Overlay Communications

Tutor: ,

Student: Reimond Retz

Completed Theses