Impact of Compressed Sensing With Quantization on UWB Receivers With Multipath Channel Estimation

Abstract

This paper explores the application of compressive sensing (CS) for ultra wide band (UWB) communication. Channel estimation is an important aspect for any communication system and especially for UWB systems in order to appropriately collect the energy from the multipath channel. UWB generally requires a high sampling rate since the bandwidth is large. Channel estimation using CS is studied along with its impact on reducing the sampling rate for an ADC to reduce power. Practical issues regarding the effect of quantization on channel estimation are addressed and a hardware implementation for CS based on the Walsh-Hadamard transform (WHT) allowing sub-Nyquist sampling is proposed. To separate the effect of channel estimation with CS, the performance of the sub-Nyquist ADC is studied in a noiseless and multipath free channel and design decisions are discussed. Comparison with the Nyquist ADC shows that using the sub-Nyquist ADC reduce power by a factor of about 6×. For the proposed hardware, two receiver architectures based on matched filtering and filtering in the compressed domain (so-called “smashed filtering”) are studied. It is found that with a perfect channel smashed filtering performs better than matched filtering. Finally the effect of channel estimation on the proposed hardware is studied along with two different recovery algorithms namely basis pursuit and matching pursuit.