Genetic Algorithm Aided Design of Near-Capacity Irregular Variable Length Codes


Maunder, R. G. and Hanzo, L. (2008) Genetic Algorithm Aided Design of Near-Capacity Irregular Variable Length Codes. In, IEEE WCNC'08, Las Vegas, Nevada, USA, 31 Mar - 03 Apr 2008. , 1256-1260.

Download

[img] PDF - Published Version
Download (201Kb)

Description/Abstract

In this paper we demonstrate that our ability to match the EXtrinsic Information Transfer (EXIT) function of an Irregular Variable Length Code (IrVLC) to that of a serially-concatenated inner code depends on the availability of a suite of component Variable Length Error Correction (VLEC) codebooks having a wide variety of inverted EXIT function shapes. We also show that the inverted EXIT function shape of a VLEC codebook depends on its coding rate and Error Correction Capability (ECC). This motivates the design of a Genetic Algorithm (GA) that searches the large VLEC parameter space to find codebooks having specific coding rates, ECCs and, hence, EXIT function shapes. The employment of this GA therefore facilitates the design of component VLEC codebook suites without the manual trial-and-error that is required when employing the state-of-the-art Heuristic Algorithm (HA) used as our bench marker, which cannot design component VLEC codebooks having specific EXIT function shapes.

Item Type: Conference or Workshop Item (Paper)
Additional Information: Event Dates: 31 March - 3 April 2008
ISBNs: 9781424419975
ISSNs: 1525-3511
Keywords: variable length coding, VLC, iterative decoding, EXIT chart matching, serial concatenation, joint source and channel coding, trellis, BCJR, soft, irregular coding, free distance, computational complexity, capacity, RVLC, VLEC, reversible variable length coding, variable length error correction coding, GA, genetic algorithm, URC, unity rate code
Divisions: Faculty of Physical Sciences and Engineering > Electronics and Computer Science > Comms, Signal Processing & Control
ePrint ID: 264589
Date Deposited: 24 Sep 2007
Last Modified: 27 Mar 2014 20:09
Further Information:Google Scholar
ISI Citation Count:0
URI: http://eprints.soton.ac.uk/id/eprint/264589

Actions (login required)

View Item View Item