Author :
Haghighat, J. ; Jamali, S.H. ; Behroozi, H. ; Nasiri-Kenari, M.
Abstract :
Recently, V. Sorokine et al. (see IEEE Trans. Commun., vol.48, no.10, p.1660-8, 2000; vol.48, no.11, p.1818-28, 2000) have introduced a class of low-density parity-check (LDPC) codes which, when applied on an IS-95 CDMA system, show, at BER=10-3, a five-fold and two-fold increase in capacity compared to the traditional uncoded scheme and to state-of-the-art low-rate orthogonal convolutional codes, respectively. However, these codes suffer from an error-floor at lower BERs which are needed for high-quality services, such as data. To enhance the capacity of CDMA systems, not only for moderate BER values but also for lower BER values, we modify the construction of these codes slightly, to improve their performance at lower BERs. We use a simple and fast iterative algorithm instead of an exhaustive search for this purpose. Our results show a three-fold increase in capacity at BER=10-5 compared to the low-rate orthogonal convolutional codes and 1.5 times increase compared to a typical code of the Sorokine et al. ensemble.
Keywords :
code division multiple access; convolutional codes; error statistics; iterative methods; parity check codes; quality of service; BER; CDMA; LDPC codes; error-floor; iterative algorithm; low-density parity-check codes; orthogonal convolutional codes; Application software; Bit error rate; Channel capacity; Convolutional codes; Decoding; Gaussian channels; Iterative algorithms; Java; Multiaccess communication; Parity check codes;