ECE 710 topic 12 - Network Information Theory
Instructor
Description
This course introduces some basic topics in multi-user information theory with an emphasis on their application to communication networks. Fundamental capacity theorems (achievability and converse) and coding strategies will be covered for both source coding and channel coding problems. Besides classical results, recent developments in the field are also covered.
Prerequisites
ECE 604, ECE 611, ECE 612 or equivalent.
Major topics
- Review of the basics of single-user information theory: source coding theorem; channel coding theorem; typical sequences; Fano's inequality.
- Multiple access channel: the capacity region. Encoding of correlated sources: Slepian-Wolf Coding.
- Broadcast channel: physically degraded, stochastically degraded, superposition coding, Marton's region.
- Relay channel: the capacity of the degraded case, block Markov coding, decode-and-forward, successive decoding, sliding-window decoding, backward decoding.
- Review with Gaussian channels: geometric interpretation, Gel'fand-Pinsker Channel and Costa's dirty paper coding.
- Rate-distortion theory and Wyner-Ziv Coding: strongly typical sequences, the CEO problem.
- Relay networks: compress-and-forward, extension to multiple relays, noisy network coding, repetitive encoding vs. cumulative encoding, joint decoding vs. successive decoding, backward decoding vs. forward decoding.
- General wireless networks: Max-flow min-cut lemma, the cut-set bound, multiple cuts, transport capacity, upper bounds of the transport capacity, relation between transport capacity and network energy, scaling laws, hierarchical cooperation schemes.
Grading
Project 30%; Final 70%