Implementation of convolutional decoding algorithms in CDMA systems (Q1276930)

The paper deals with some selected aspects of CDMA systems operation. The authors consider and compare effectiveness of two encoding and decoding strategies influencing the complexity of the CDMA transmitter and receiver structures. Instead of the encoding method used in the Qualcomm solution they propose a new method leading to better performance parameters. The original method assumes that after convolutional encoding, repetition, and interleaving, binary symbols are arranged into blocks of \(k\) symbols. Each of these blocks is mapped into a corresponding row of a \(q*q\) Hadamard-Walsh (HW) matrix. Then the code sequences are multiplied symbol-by-symbol by long pseudonoise (PN) sequences (playing the roles of user addresses). On reception, the receiver correlates the received signals with his own PN sequence and with all \(q\) possible rows of the HM matrix. The decoder chooses a row of the HM matrix with the largest correlation function. Then binary decoding is performed. Contrary to the above method the authors develop another approach. They use the same technique for HW mapping, but instead of binary encoding they use \(q\)-ary convolutional encoding. Then they perform PN address transformation of \(q\)-ary symbols before the HW mapping. Thus, unlike in the original Qualcomm scheme the receiver is able to distinguish all transmitted \(q\)-ary symbols at each time instant. This procedure is performed by the conventional bank of \(q\) correlators and some threshold selector. In the proposed method the decoding process is reduced to the sequential analysis of the code tree nodes. For this purpose the authors develop a stack algorithm. The simulation results show that the new method significantly improves the CDMA system operation as compared to the original Qualcomm system.