Vector Symbol Decoding for Systematic Nonbinary Convolutional Codes in Narrowband Power Line Communications

摘要

Nonsystematic convolutional (conv.) codes can provide higher constraint length and free distance, which lead to better performance than the systematic codes. However, for large nonbinary symbols that cannot be practically decoded by Viterbi decoder and need to be decoded with Vector Symbol Decoding (VSD) technique instead, using systematic conv. codes have two main advantages. First, the method of calculating the parity check matrix is greatly simplified. Second, the decoded data can be directly read from the decoded sequence with no additional circuit to map them. Thus, Bit error rate (BER) can also be found in addition to the decoding failure probability usually shown in previous works. Results show that with (63,47) Reed-Solomon inner - (3,2,2) systematic conv. outer code, BER is 1.56×10~(-5) in the Narrowband Power Line Communications (NB-PLC) channel modeled by the Middleton Class A with A=0.1 gamma= 0.0001and AWGN with SNR= 8 dB. With (63, 39) RS and (63, 33) RS, the BERs are 9.23x10~(-7) and 5.3824 ×10~(-7) for SNR =6dB respectively.