| CDMA assigns a distinct signature code to every user so as to separate their signals. As a result, the mobile communication systems employing this multiple access technique have an advantage of large capacity, good security and wide coverage area etc., however, due to the nonideal orthogonality between the signature codes and different transmission delays, when user signals arrive at base station multiple access interference (MAI) will emerge between them inevitably.In order to decrease MAI, multiuser detection (MUD) has been proposed and parallel interference cancellation (PIC) is an effective MUD method. Since the detection capability of PIC lies on the decision veracity at each stage, inaccurate decision value from the former stage will result in false interference cancellation at the latter stage, which will affect sequent decision. To solve this problem, partial parallel interference cancellation (PPIC) was proposed, i. e., multiply the regenerated MAI by a coefficient between 0 and 1 which is called interference cancellation factor (ICF).Many kinds of algorithms have been proposed to generate ICF. Based on analysis of the factors which affect the veracity of decision, this paper proposed two new PPIC algorithms, partial parallel interference cancellation based on Hebb learning rule (Hebb-PPIC) and fuzzy parallel interference cancellation based on estimated signal-to-interference ratio (SIR-FPIC), and consequently tested their detection capability via computer simulation. In detail, this paper mainly consists of three parts:(1) The theories of two kinds of communication techniques, DS-CDMA and MC-CDMA, were given first, on the basis of which the two corresponding system models were established. A brief introduction to the origin, development and classification of MUD was presented later. Preparing for the proposal of new algorithms then put emphasis on the introduction to matched filter detection, PIC and PPIC.(2) Summarized the origin and development of artificial neural network and then described the fundamental of Hebb learning rule. Based on this, this paper presented the structure and interference cancellation process of Hebb-PPIC. In order to investigate the detection capability of Hebb-PPIC, consequently a scries of simulations were performed for ideal power control and near-far scenario in DS-CDMA and MC-CDMA system models via MATLAB software. The simulation results showed that compared with conventional PIC Hebb-PPIC could be able to lower the bit error rate (BER) and improve the detection capability effectively.(3) Considering that the ICF in Hebb-PPIC is not a real-time parameter, the SIR-FPIC algorithm was proposed that apply the estimated SIR of the former stage to constructing a membership function as the ICF of the latter stage. In the same conditions corresponding simulations were performed whose results indicated that compared with Hebb-PPIC and PIC SIR-FPIC could dramatically lower the reception BER, which proved that SIR-FPIC had successfully overcome the shortcomings of Hebb-PPIC. |
