EE209AS Spring 2011 Prof. Danijela Cabric Paper Presentation Presented by: Sina Basir-Kazeruni sinabk@ucla.edu A Subsampling UWB Radio Architecture By Analytic Signaling by Mike S. W. Chen and Robert W. Brodersen
Introduction Main Goal: Improve the efficiency of UWB implementation using signal processing techniques. System architecture, implementation challenges are presented. Digital analytic signal processing is discussed. 2
Freq response Transmitter [1] I Q 90 shift Bp Bp Narrowband System: A baseband signal is mixed up to carrier frequency UWB system: Pulser drives antenna and generates a passband signal without mixer
Receiver UWB attenna Bp LNA A D C Digital Backend [1] Incoming signal directly sampled after amplification. Sampled data processed by a digital matched filter To reach the matched filter bound for optimal detection. Avoids wideband analog processing Adds more processing to the digital backend. 4
Freq response Freq response Subsampling Subsampling samples the passband signal at twice the signal bandwidth instead of the maximum signal frequency. Signal is bandlimited from Fl to Fh and the sampling frequency is Fs. By carefully choosing Fs, Fl and Fh a non-aliased sampled spectrum can be derived: F l = k. (F h - F l ) = k. B, where k N k is undersampling ratio 5 -π Fs π B Fl Fh [1]
Subsampling Subsampling Challenges: Noise spectrum from Fl to +Fl will alias into the signal band. Aliasing is proportional to fc/b. More difficult to design a bandpass filter in RF than IF or baseband. Sampling jitter causes severer degradation of the SNR. Directly sampling at RF frequency introduces more noise power than IF or baseband frequency. Most of this challenges exist for narrowband and become reasonable for UWB. UWB can suffer from sampling offset (timing sensitivity) Frequency mismatch between the TX and RX oscillators or changes of pulse arrival times. 6
Impact of Sampling Offset in UWB Impact of sampling offset is shown below for measured UWB pulses that are bandlimited to 3-4GHz. The proposed analytic signaling approach is implemented to alleviate this problem. 7
Analytic Signal Processing Bandlimited signal, s(t), is sampled at 1/Ts. Any sampling offset, To, of the sampling sequence will transform into a phase shift in frequency domain: Approach 1: throw away the phase term by calculating the magnitude of signal s FFT. 8 Not optimal Loss of phase information Approach 2: Analytic Signal Processing. Energy detection can be decoupled from the phase shift, if we formulate an analytic signal. Real and imaginary part of the analytic signal are orthogonal. As sampling offset varies, signal energy moves between these two orthogonal dimensions.
Digital Backend The main components are: Pulse shape estimator Analytic signal transformer Correlators Analytic matched filter Detection block. The analytic matched filter response is: Analytic matched filter takes the pulse estimator results and convolves it with the incoming analytic signal. The correlation block is used to provide additional processing gain or despread any possible coding that is modulated on the pulses. Detection block is used for synchronization and data recovery. 9
Performance Evaluation Performance Comparison of Real-Valued and Analytic Signal Processing Vulnerability to the timing offset is reduced through analytic signal processing The limitations of using subsampling frontend are alleviated. 10
Performance Evaluation Usage of Analytic Signal for Timing Information Timing offset can be caused by pulse delay Utilize this to measure the distance between TX and RX. For ranging purpose, high sensitivity to timing offset implies a high time resolution for the system. The higher the frequency band that UWB pulses use, the finer the time resolution will be for the system Accuracy of ranging also depends on how fine the system can resolve the angle. Directly related to SNR of analytic matched filter output. 11
Additional Reference [1] Mike Chen. "A Subsampling Radio Architecture for 3-10 GHz UWB". Talk or presentation, BWRC Summer 2003 Retreat, 11, June, 2003. 12
Questions? 13