Version 1

1
TESTING OF P.1812 DIFFRACTION MODELS WITH RESPECT
TO MOBILE MEASUREMENT DATA
Richard Rudd 3K1 correspondence group
workshop London, 11-13 March 2008
2
Background

• Testing of diffraction models against mobile
  measurements
• Majority of P.1812 testing relates to paths of
  gt10km and fixed receivers, typically 10m height
• Recent OFCOM-funded mobile measurements at 2-6
  GHz
• Opportunity taken to examine performance of
  Deygout three-edge and Bullington models with
  respect to this data
• Caveats
• Majority of measurement data is outside intended
  frequency range of P.1812
• Is P.1812 intended to be used for such short
  paths?

3
Measurement campaign

• Both fixed and mobile measurements
• Frequencies of
• 2.4 GHz
• 3.4 GHz
• 5.7 GHz
• Total of 85 measurement runs

4
Measurement areas

• Measurements in four UK areas
• Suburban Cheshire (fixed only)
• Rural Oxfordshire
• Winchester (small cathedral city)
• Croydon (dense urban area in London)

5
9th order distance correction

• Bullington implemented as per CGD-12
• Includes taper to ensure free-space for LOS paths
• 9th-order distance correction as per CGD-14,
  applied
• Additively, to CGD-12 algorithm
• As a substitute for existing distance term

6
Overall results
(But D3E as good as Bullington if related edges
identified)
7
Assumed clutter heights
8
Sensitivity to assumed clutter height

• Tuning clutter heights improves the model
  performance
• Not tuned for these models, but optimisation
  likely to be possible

9
Rural route (2.4 GHz)
10
Winchester (5.7 GHz)
11
Undulating rural route (5.7 GHz)
12
Profile spacing
13
Comments

• These comparisons may not be relevant to P.1812
• Frequencies largely outside model range
• Paths (possibly) much shorter than intended
  applications
• Bullington performance is good
• Avoids spurious edge issues with no apparent loss
  of detail
• But other comparisons show Bullington to fail on
  longer paths
• Systematic understimate of spherical surface loss
• Deygout 3E gives realistic answers by
  non-physical means
• Making Bullington look like D3E at longer
  distances may be one answer
• Or one might correct D3E to give better
  performance at short range

14
Issues

• Not much data available for shorter, mobile paths
• Clutter information a particular difficulty
• Need to tune representative clutter heights
• Should this type of path be covered by P.1812?
• Generally, what is the specification for P.1812
• Is an empirical correction for long (spherical
  diffraction) paths the best approach?
• Polynomial or other correction to Bullington?
• accidental fit of D3E?

15
Thank you!
Richard Rudd Aegis Systems Limited www.aegis-sys
tems.co.uk