Low-Complexity Equalization Based on Least Squares Support Vector Classifiers for DS-UWB Systems

We propose the least squares support vector classifier (LS-SVC) based equalization schemes for direct sequence ultra wideband (DS-UWB) systems, where a bank of independent LS-SVCs are employed to detect each block of signals. The LS-SVC based equalizers provide a close bit error rate (BER) performance in the line-of-sight (LOS) scenario to the case with additive white Gaussian noise (AWGN). Simulation results show that the LS-SVC based equalizers have almost identical BER performance to that of typical support vector classifiers (SVCs) with a reduced training complexity. Furthermore, the sparse LS- SVCs are employed to reduce the detection complexity, with little performance loss compared to LS-SVCs.