A reduced order code spectrum transform using auto-regression (ROCSTAR) for fast, low FAR CDMA forward link code estimation

In the CDMA synchronous forward link, the mobile only has knowledge of its own spreading code; therefore, blind interference suppression (IS) algorithms are required to mitigate the multiple access interference (MAI). In this paper, an algorithm using reduced rank auto-regression (AR) methods is introduced to allow the mobile to determine all of the codes that the base station is transmitting. The full rank AR technique is based on computing an orthogonal "code spectrum", analogous to the traditional Fourier spectrum for frequency analysis. The reduced rank technique, called reduced order code spectrum transform using auto-regression (ROCSTAR), greatly improves performance by eliminating spurious peaks in the code spectrum and is based on the multistage Wiener filter (MWF). It is shown that this technique is more efficient, provides better probability of detection and lower false alarm rate, and operates with fewer samples than the full rank solution. This robust method of code estimation allows interference cancellation (IC) to be performed under severe channel conditions typical of dense urban environments where existing IS methods fail. Performance of an MWF based IC algorithm using ROCSTAR is evaluated and compared to the conventional minimum mean square error (MMSE) receiver and blind MWF receiver, which has previously been shown to outperform all other receivers. It is shown that with few samples, at least 10 dB gains over the blind MWF and the MMSE receiver is achieved. This opens the door to the development of new algorithms for channel estimation and IC. This technique is expected to have tremendous impact on the performance and capacity of future CDMA systems