Joint Decoding Of LDPC Code And PTS Phase Factors For OFDM Systems

One major drawback of orthogonal frequency-division multiplexing (OFDM)systems is the high peak-to-average power ratio (PAPR). When OFDM signals with peakpower passing the high power amplifiers (HPA), they would distort the in-band signaland raise the out-of-band radiation. Partial transmit sequence (PTS) technique cansignificantly reduce the PAPR without introducing distortion to the transmitted signal.However, one of critical challenges of PTS schemes is that the phase factor informationis required to be transmitted to receiver as side information, which decreases thetransmission efficiency and increases the complexity of the system design.First, we propose joint decoding of the LDPC code and phase factors. By adoptingthe proposed joint decoding scheme in an LDPC-coded OFDM system with PTS PAPRreduction, we could avoid transmitting PTS side information about the phase factors. Thekey idea is that we view the PTS processing as a stage of coding, and call the resultedcode of LDPC coding and PTS processing as concatenated LDPC-PTS code. Then, wederive the parity-check matrix of the concatenated LDPC-PTS code. With theparity-check matrix, the LDPC code and phase factors can be jointly decoded using beliefpropagation (BP) algorithms with nearly perfect performance.Then, in order to enhance the decoding convergence speed of the concatenatedLDPC-PTS code and support large number of PTS partitions, we proposed an optimizedpartition method based on greedy algorithm. The key idea is to minimize the degree ofphase node in the extended Tanner graph. Particularly, the partition of OFDM symbols isfirst equivalent to the partition of columns in the parity-check matrix. Then, whenassigning a new column for a sub-matrix, it should guarantee that the degree of the current phase node corresponding to the sub-matrix should be minimized. On the basis ofthe greedy algorithm, we further propose an iterative greedy algorithm to reduce thedegree of phase nodes.