Simulating social-based forwarding in opportunistic networks

   page       attach   
Luca Mella
abstract

The background theme of this project is the social-based forwarding in opportunistic network, so, broadly speaking, we are facing the problem of delivery or acquire a content in an infrastructure-free scenario composed by a multitude of mobile-device (typically carried by humans) by exploiting what we know about social relationship between devices/humans with respect to their movements and their consequential physical contacts. This class of problems were one of the topics in SOCIALNETS: Social networking for pervasive adaptation, an European FP7-ICT research project which investigated this kind of problems and produced several prototypes spendable in infrastructure-free scenario.

In particular, we are focusing in a subtopic treated during this project: algorithms for social-based forwarding in opportunistic, delay tolerant network. Several algorithms have been evaluated during the course of SOCIALNETS and an interesting fact we have noted is that performed experiments were mainly based on emulation approaches, so due to the notorious difficulty of collecting this kind of data, in fact collecting process involved lot of people and required from days to years, we consider interesting to experiment this kind of algorithm in a simulated environment, trying to adopt proper mobility models in order to obtain realistic performance measurement.

The reference paper for this class project is BUBBLE Rap: Social-based Forwarding in Delay Tolerant Networks 

outcomes
final report

Goals

We want to measure efficiency and efficacy of social-based forwarding algorithm through a different approach based on ABM simulations. so:
  • First we need to set up an simulation environment which enable us to simulate credible human mobility pattern driven by the social relationship between mobile users.
  • Second we have to set up experimental sessions in order to measure delivery ratio and delivery cost of several forwarding algorithms (both social-based and not).
  • Then we will have to discuss about obtained results and perform a comparison with the outcome pointed by our reference paper.

Requirements

We need a simulation environment which supports both mobility and social networking concepts.

Simulation Platform

Our experiment mainly involves two main concepts: mobile nodes and social networks. Since we think supporting mobility simulation is generally harder than supporting social network simulations we choose to adopt the Alchemist simulation platform because it has been explicitly designed to handle nodes mobility and contact networks.

Work Plan

Considering our technological choice (driven by a review of several simulation platforms) we have subdivided our work as follow:
  1. Document about social network structures, characteristics and models, about mobility models in general and mobility models which exploits social network topologies, and also forwarding algorithm in opportunistic networks.
  2. Add to Alchemist the concept of social network and enable simulation on arbitrary topologies.
  3. Design a mechanism which permit to define arbitrary network formation algorithm
  4. Support environment with multiple networking layers. For instance support an euclidean environment with mobile nodes that links each others based on their physical distance, and at the same time support the concept social neighbourhood, or in general another linking layer based on other - non euclidean - criteria.
  5. Run experimental sessions and measure delivery ratio and delivery cost.
  6. Perform analysis and statistics on collected data, then compare analysis outcomes with the reference paper ones.

Outcome

We point to the class project report for a detailed description of the work done and the analysis of experimental results:Report SMA1213 - Simulating social-based forwarding in opportunistic networksThe Alchemist extension developed during the project is publicly available at: Alchemist SocialnetsAnd the collected data are also available at:Simulation data -part001 Simulation data -part002 Simulation data -part003 Simulation data -part004