A hybrid multiuser detector for DS-UWB systems with matched filter and error-bit recognizer

In this article, we propose a hybrid algorithm of multiuser detection (MUD) for direct sequence ultra-wideband (UWB) communication systems, based on matched filters (MF) and error-bit recognizers (ER). The ER embraces three steps: bit mapping, K-means clustering, and sign judgment. In this detector, we construct a mapping function and map the output base-band bits from MFs into a one-dimensional feature space to get the amplitude information. After that, we set an initial value and classify the bits into right bits and wrong ones by K-means clustering. Finally, we make the sign judgment to prevent some right bits being picked out as wrong bits and correct the wrong bits to reduce the bit error rate (BER) of the system. Compared to a conventional detector based on MFs and minimum mean square error detector, to a large extent, our detector can improve the BER performance of a system because of high recognition probability of the wrong bits in K-means clustering. What is more, the simulation results show that the proposed hybrid MUD is not sensitive to the number of users. It can achieve a good BER performance and low computational complexity when there are 20 UWB transmitters in the channel.