Concept of Subsampling

1
Concept of Subsampling

• Begin with a bandlimited signal at a certain
  frequency band Fl Fh

B

• Receiver samples at Fs, the resulting signal
  spectrum is the repetitive replicas of integer
  multiple fs

p
-p
2
Integrated Analog Front End

• Radio Architecture
• - Wideband antenna
• - Wideband amplifier / matching network
• - RF bandpass filtering (low Q filter)
• - High bandwidth sample and track
• - High-speed and low resolution ADC
• - Sampling Clock generator
• - DSP

3
How to recover the information?

• Reach the matched filter bound digitally without
  wideband analog processing. Another good example
  of taking the advantage of technology scaling.
• How to know the matched filter response?
• A Simply use running average during training
  sequence. The best linear estimator, if the noise
  is AWGN.
• A small shift of sampling time will change the
  sampled waveform dramatically. A timing recovery
  issue! Matched filtering failed
• A Propose a Complex signal based digital
  backend

4
Analytic Signal Processing

• Convert a discrete real signal into a discrete
  analytic signal via Hilbert transformer.
• Real and imaginary part of the analytic signal
  are two orthogonal sets. The analogy is I and Q
  channel!
• Analytic MF response h(t) s(T-t).

Extract the hidden info by analytic signaling!
5
Constellation _at_ Analytic MF outputs

• Timing offset cause the constellation to rotate.

Imag
0Ts
0.05Ts
signalnoise
noise
Real
0.1Ts
0.15Ts
6
Trajectory with infinite-bit ADC

• Sampling offset varies from 0.1Ts to 4Ts.
• SNR of Real, Imag, Abs of analytic MF output.
• Timing Information!

7
Trajectory with 1-bit ADC

• More averaging is needed or the robustness to the
  sampling offset is reduced.

8
Subsampling Analytic Filtering

• Possible application
• Data communication or Ranging at a low cost

9
Conclusion

• A subsampling analog front end combined with
  analytic signal processing has been proposed for
  passband UWB communication.
• By the magnitude of complex MF output, timing
  recovery has been done without interpolation or
  oversampling. Thus, it makes digital matched
  filtering promising and practical.
• This technique also achieves a very fine timing
  resolution smaller than sampling rate, which
  implies high accuracy ranging capability. This is
  a special feature of UWB.