The performances of convolutional codes used in turbo codes [articol]

Abstract

In this paper are presented and compared the BER performances obtained by the simulation of a transmission system, which utilizes the forward error correcting by codes concatenation and iterative decoding (turbo coding). There have been investigated all the systematic convolutional codes having the constraint length K less or equal to 6, under three diffrent concatenated forms: parallel PCCC (pure turbo code), serial SCCC and hybrid HCCC, at the following rates: 1/3, 1/4 and 1/3, respectively, all unpunctured. Two interleaver types were used: pseudo-random and S-interleaver having the same length, 1784. The AWGN channel and the BPSK modulation were employed. The used iteration number was eight. For increasing the work speed an iterations stop criterion was used. When the resulting error number from the decoding of a data block is zero, the remaining iterations are not effectuated, passing to the next block. For decoding, the MAP, MaxLogMAP and Log MAP algorithms were used. In all the cases, a tail off was employed for the first code, with the decreasing transmission rate price. The transmitted data block numbers for a simulation were chosen in function of the signal to noise ratio, SNR, i.e. to keep a good precision for obtained curve.