This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Error Correcting Codes: Combinatorics, Algorithms and Applications (Fall 2007) Lecture 16: Plotkin Bound October 2, 2007 Lecturer: Atri Rudra Scribe: Nathan Russell & Atri Rudra In the last lecture we proved the GV bound, which states that for all δ with ≤ δ ≤ 1 − 1 q , there exists a qary code of distance δ and rate at least 1 − H q ( δ ) − ε , for every ε > . In fact we proved that with high probability, a random linear code C lies on the GV bound. We picked a generator matrix G at random and proved the latter result. At this point, we might ask what happens if G does not have full rank? There are two ways to deal with this. First, we can show that with high probability G does have full rank, so that  C  = q k . However, the proof from last lecture already showed that, with high probability, the distance is greater than zero, which implies that distinct messages will be mapped to distinct codewords and thus  C  = q k . Further, the proof required that δ ≤ 1 − 1 q because it is needed for the volume bound – V ol q ( , δn ) ≤ q H q ( δ ) n – to hold. It is natural to wonder if the above is just an artifact of the proof or, for example, is it possible to get R > and δ > 1 − 1 q ? In today’s lecture, we will show that this cannot be the case by proving the Plotkin bound. 1 Plotkin Bound We start by stating the Plotkin bound....
View
Full
Document
This note was uploaded on 01/02/2012 for the course CSE 545 taught by Professor Rudra during the Spring '11 term at SUNY Buffalo.
 Spring '11
 RUDRA
 Algorithms

Click to edit the document details