Title :
Local graph coloring and index coding
Author :
Shanmugam, Karthikeyan ; Dimakis, Alexandros G. ; Langberg, Michael
Author_Institution :
Dept. of Electr. & Comput. Eng., Univ. of Texas at Austin, Austin, TX, USA
Abstract :
We present a novel upper bound for the optimal index coding rate. Our bound uses a graph theoretic quantity called the local chromatic number. We show how a good local coloring can be used to create a good index code. The local coloring is used as an alignment guide to assign index coding vectors from a general position MDS code. We further show that a natural LP relaxation yields an even stronger index code. Our bounds provably outperform the state of the art on index coding but at most by a constant factor.
Keywords :
encoding; graph colouring; general position MDS code; graph theoretic quantity; index code; index coding vectors; local chromatic number; local graph coloring; natural LP relaxation yields; optimal index coding rate; Channel coding; Color; Indexes; Interference; Vectors;
Conference_Titel :
Information Theory Proceedings (ISIT), 2013 IEEE International Symposium on
Conference_Location :
Istanbul
DOI :
10.1109/ISIT.2013.6620407