Lattice codes for Gaussian networks

Contact: Raj Kumar
Room: BC 160
Tel: 021-693-6791
Email: raj [dot] kumar [at] epfl [dot] ch

Determining the capacity of wireless networks remains a “holy grail” of multi-user Information theory, but several recent advances have been made in this regard. While most traditional achievability results in information theory have relied upon the tool of random coding, recent results have shown several interesting cases where structured codes such as linear and lattice codes outperform purely random codes.

The objective of this project is to:

  • Understand the use of lattice codes for communicating over Gaussian networks
  • Extend a recent result showing that lattice coding is optimal for a particular class of Gaussian relay networks to more general families of networks
  • Explore connections with the recently proposed deterministic channel model

This project will be suitable to a student with a theoretical inclination.

Prerequisites: Solid background in Information Theory and Mathematics.