Schmidl-Cox-like Frequency Offset Estimation in Time-Hopping UWB

This paper presents a time hopping ultra wide band (TH-UWB) receiver design targeted to high performance and low complexity. The classical Schmidl and Cox idea of extracting the frequency offset from the phase of a correlation measure between two identical transmitted signals is here extended to a TH format. The algorithm exploits the low duty cycle of the time hopping access, and combines received samples in order to strengthen the signal-to-noise ratio. A clever selection of which samples to use in the correlation measure (first and last third of the received signal) is proved to perform 0.5 dB from the Cramer-Rao lower bound, thus providing an improvement of 0.75 dB with respect to the standard approach of the literature (correlating the first and second halves of the received signal). A modified algorithm that suitably weights received samples is also proposed for achieving robustness in impulsive multiple user access scenarios.