Doppler Spectrum Measurement of HF Channel Simulators

1
Doppler Spectrum Measurement of HF Channel
Simulators

• William N. Furman, Harris Fellow, Senior
  Scientist
• John W. Nieto, Senior Scientist
• Harris Corporation
• RF Communications Division

2
Introduction

• Previous HFIA presentations and conference papers
  have discussed the importance of HF channel
  simulator characteristics 12
• One channel simulator attribute that has been
  shown to have a large impact on measured results
  is the spectrum of the fading channel vector
• The generally accepted criteria is that this
  spectrum should be Gaussian in shape
• This presentation documents a method to evaluate
  the fading spectrum of a channel simulator under
  test as a black box

3
Overview

• Watterson Channel Model
• Generation of Fading Vector
• Importance of Fading Spectrum
• Proposed Measurement Technique
• Conclusions

4
Watterson Channel Model

• Model proposed in 1970, experimentally validated.
• Stationary, wrt. Doppler Spread and SNR, model
  valid for short durations (10 minutes) and for
  band-limited channels ( 10kHz).
• Models the channel as a transversal filter where
  the taps are complex and vary with time. Includes
  the addition of band limited additive White
  Gaussian Noise.

5
Watterson Channel Model
6
Generation Of Fading Vector

• Fading filter taps generated by low pass
  filtering of independent Gaussian random
  variables.
• Magnitude of fading vector will be Rayleigh
  distributed.
• Characteristics of this fading vector generation,
  such as filter shape, interpolation, and update
  rate are all very important considerations in
  obtaining repeatable performance measurements.

7
Example Spectra (IDEAL)
8
Importance Of Fading Spectrum
9
Proposed Measurement Technique

• Step 1 - Set up HF channel simulator for 1 path
  with a Doppler spread of X Hz.
• Step 2 - Pass a 1 KHz tone through the HF
  channel simulator recording the output with a
  high quality DAT recorder. (44,100 sps)
• Recommend 2 hour recording

10
Proposed Measurement Technique

• Step 3 - Input DAT samples into PC WAV file
• Recommend using optical or coaxial SPDIF transfer
  for bit perfect transfer of digital data on DAT
  tape.
• Warning not all soundcards with digital inputs
  give you a bit perfect transfer. We utilize the
  M-Audio stereophile 2496. It does.
  http//www.midiman.net/products/m-audio/audiophile
  .php
• Step 4 - Convert WAV file from 44,100 sps to
  11,025 sps (mono)
• Steps 1-4 result in a high quality digital
  recording of the output of the channel simulator
  under test. Be wary of standard sound cards and
  converters.

11
Proposed Measurement Technique

• Step 5 - Process WAV File
• Take 131,072 point FFTs of frames of the sample
  stream
• Take the magnitude-squared of the FFT output
• Average the magnitude-squared output for all
  frames
• Plot spectrum. Plot is an estimate subject to
  spectral analysis effects. ( sinx/x windowing)
• Step 6 - Check resulting spectrum
• Compare to ideal Gaussian Shape.
• Close agreement from 0 to 20dB
• Method described in 2 details how to generate
  the Gaussian spectrum

12
Measured Spectra
13
Improved Specification
14
Conclusions

• The fading process generation on an HF channel
  simulator implementation can result in measurable
  performance deviations at 2400 bps and
  significant performance deviations at 9600bps.
• This presentation has detailed a measurement
  technique that will provide a mechanism for easy
  evaluation of the Doppler spectrum of an HF
  channel simulator and will hopefully minimize
  simulator to simulator measurement variations.

15
References

• 1 - Requirements for HF Channel Simulators
  Fading Characteristics. HFIA Presentation June
  2000. W.Furman J. Nieto. Harris Corporation.
• 2 - Understanding HF channel simulator
  requirements in order to reduce HF modem
  performance measurement variability. Proceedings
  of HF01, the Nordic HF Conference, August 2001.
  W.Furman and J. Nieto. Harris Corporation.