Assignment: cross layer learning

Finding the Truth:
Probabilistic Graphical Models for cross-layer inference  

Problem Statement: 

Very often the effects of lower layer parameters (e.g. Transmit power at PHY or Header size at MAC) have strong effects on the application as well. These are generally difficult to model. Probabilistic Graphical Models are a very powerful tool able to express the inter-dependencies between variables, such as 


The students will receive measurement data from realistic Wireless Sensor Network and 

  1. design the PGM (in MATLAB or any other language) to determine the interdependencies between PHY, MAC and APPLICATION variables, 
  2. determine solutions based on the PGM to solve specific tasks (i.e.: reduce errors, or increase reliability) 


40% Theory, 40% Simulations, 20%Writing.


Alessandro Chiumento (