DocumentCode :
2286193
Title :
Evolutionary coded modulation: a coded-modulation extension of turbo codes designed with genetic algorithms
Author :
Snelling, W.E.
Author_Institution :
Inventive Res. Inc., Columbia, MD, USA
Volume :
2
fYear :
2001
fDate :
2001
Firstpage :
1133
Abstract :
Evolutionary coded modulation targets the realm of small block sizes or high rates; a realm where parallel, concatenated turbo codes (PCC) followed by antipodal modulation is comparably inefficient. This inefficiency stems mainly from termination overhead and from the discrepancy in channel capacity between binary and continuously valued signals. This paper reports two complementary but independent results that comprise evolutionary coded modulation. The first result is a class of coded modulation that is a near superclass of the following; PCC codes with antipodal modulation, trellis coded turbo modulation, turbo-coded modulation and superorthogonal turbo-codes. In contrast, evolutionary coded modulation features these: subcode output coupling - PCC expect subcode output bits to be independently transferred through the channel. Our method allows subcode output bits to be arbitrary coupled. Nonsystematic coding PCC rely on direct measurement of the information bits from the channel. Our method does not require or benefit from systematic coding. Nonlinear coding - PCC consist of linear subcodes and are linear overall. Our method is nonlinear both at the code and modulation levels. The second result comprising evolutionary coded modulation is a way to find near optimal members of the class using genetic algorithms. By applying judicious constraints, we find a genetic encoding that eliminates most catastrophic members but still admits near optimal members. Also key is a way of calculating the fitness of candidate members that is more computationally efficient than computing bit error rate. The performance of evolutionary coded modulation over PCC codes with antipodal modulation is evaluated. For blocks of 256 bits, a 0.5 dB gain is reported. Of that approximately 0.05 dB stems from a more optimal decoding algorithm, about 0.13 dB comes from lower termination overhead and the remainder comes from using 16 QAM in place of antipodal modulation.
Keywords :
concatenated codes; genetic algorithms; iterative decoding; linear codes; modulation coding; nonlinear codes; quadrature amplitude modulation; turbo codes; 16 QAM; antipodal modulation; binary signals; bit error rate; channel capacity; continuously valued signals; evolutionary coded modulation; gain; genetic algorithms; iterative decoding; linear subcodes; modulation levels; nonlinear coding; nonsystematic coding; optimal decoding algorithm; parallel concatenated turbo codes; subcode output coupling; superorthogonal turbo-codes; systematic coding; termination overhead; trellis coded turbo modulation; turbo-coded modulation; Algorithm design and analysis; Bit error rate; Channel capacity; Concatenated codes; Decoding; Gain; Genetic algorithms; Modulation coding; Quadrature amplitude modulation; Turbo codes;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Military Communications Conference, 2001. MILCOM 2001. Communications for Network-Centric Operations: Creating the Information Force. IEEE
Print_ISBN :
0-7803-7225-5
Type :
conf
DOI :
10.1109/MILCOM.2001.986023
Filename :
986023
Link To Document :
بازگشت