Blind Estimation of Multiuser Long-Code DSSS Signals via Subspace Reconstruction and ILSP

This paper investigates the blind estimation problem of multiuser long-code Direct Sequence Spread Spectrum (DSSS) signals in non-cooperative reception scenarios. In a long-code DSSS system, the spreading-sequence length exceeds the spreading gain and is not an integer multiple of it, so each symbol interval contains only a fragment of the spreading sequence. Consequently, conventional short-code subspace methods cannot directly estimate the complete spreading-code subspace. To address this issue, a blind despreading method based on signal subspace reconstruction and iterative least-squares projection (ILSP) is proposed. First, symbol samples sharing identical spreading-sequence fragments are grouped, and subspace decomposition is applied to obtain multiple fragment subspace estimates. These fragment subspaces are then jointly combined through the periodic selection structure to reconstruct the complete spreading-code subspace. Finally, ILSP is employed to estimate the spreading sequences and recover the transmitted information sequences. Simulation results demonstrate that the proposed method remains effective under low signal-to-noise ratio (SNR) conditions and that its information-sequence bit error rate (BER) approaches that of cooperative reception as the SNR or sample size increases.