Research On Blind Adaptive Multi-user Detection

The code division multi-access (CDMA) will cause multi-access interference (MAI). Blind multi-user detection can excellently suppress the MAI with other interfere signal, which don't need the training chips and prior knowledge of interferential users. But the current algorithms need large computation and convergent slowly. They can't satisfy the real-time request. This paper proposes a signal subspace deflation tracking of NIC rule (NICD) and a scheme which combines the NICD and the traditional blind MMSE detector of signal subspace.Compared to the current blind multi-user detectors, this detector has several merits below:(1) The NICD utilizes the logarithm to project the quadratic searching process and the learning step size is adaptive. It will help convergence faster to the basis of the signal subspace.(2) The algorithm can automatically make the eigenvectors orthogonal in the iteration, which avoid the extra the orthogonal constraint and contribute to the low complexion. The orthogonal of the subspace decrease the tracking error track and lead faster convergence.(3) An improved dimension estimation technique with Akaike information criterion (AIC) is adopted during the subspace tracking. Other estimation technique need compute all the AIC value which is in direct proportion to the dimension. According the analysis of AIC function, the improved way only compute the 3 AIC values in the iteration.The performance analysis of this detector and the simulation results indicate that the blind adaptive multi-user detector proposed in this paper shows superior detection performance, low complexion, excellent anti-interference of MAI and Near-Far effect.