Packet Detection in a Filter Bank-Based Ultra-Wideband Communication System

Recently, filter bank multi-carrier spread spectrum (FBMC-SS) technology has been proposed for use in ultra-wideband (UWB) communication systems. It has been noted that, due to the spectral partitioning properties of the filter banks, a UWB signal can be synthesized and processed using a parallel set of signal processors operating at a moderate rate. This transceiver architecture can be used to generate UWB signals, without requiring a high-rate analog-to-digital and/or digital-to-analog converter. In this paper, beginning with a design operating on a single signal processor, we explore the development of a packet detector using the Rao score test. Taking advantage of the FBMC-SS signal structure, an effective detector design based on a cascade channelizer is proposed. We refer to this design as singe-radio band (SRB) detector. Given the typical bandwidth of UWB systems ($\bf 500$~MHz or wider), the SRB detector has to operate at a fast sampling rate of greater than $\bf 500$~MHz. This may be undesirable, as low cost analog-to-digital (ADC) and digital-to-analog (DAC) converters are often limited to a sampling rate of $\bf 200$~MHz or lower. Taking note of this point, the proposed SRB detector is extended to a multi-radio band (MRB) detector, where a set of parallel signal processors operating at a moderate sampling rate are used for a more practical implementation of the detector. Through computer simulations, we show that SRB and MRB detectors have the same performance in typical UWB channels. Finally, we provide results from an over-the-air demonstration of a UWB design occupying $\bf 1.28$~GHz of bandwidth. We find that reliable detection performance is possible in the harshest environments, at signal-to-noise ratios as low as $\bf -40$~dB with a preamble length of approximately half the duration of longest preamble length recommended in the IEEE802.15.4 standard.