Page 37 - ITU Journal: Volume 2, No. 1 - Special issue - Propagation modelling for advanced future radio systems - Challenges for a congested radio spectrum
P. 37

ITU Journal: ICT Discoveries, Vol. 2(1), December 2019



          is  acting  as  a  filter  that  preserves  the  dominant   narrower  6°  beam.  The  reduction  factor  between
          direct path but removes most indirect paths, thus n   the LoS istotropic and 20°-beam-width values is 4.
          is almost 2. With  larger antenna  beam width,  the   Then  the  LoS  median  delay  spread  with  6°  beam
          canyoning  effect  (or  sum  of  multiple  propagation   width goes to 0. The variation versus frequency is
          paths  caused  by  the  strong  reflections  in  this   shown  in  Fig.  8.  The  LoS  delay  spread  is  almost
          confined environment) leads to a  received power     constant  with  frequency,  meaning  the  major
          greater or equal to the LoS direct path power; n is   indirect  paths  remain  of  the  same  relative
          decreasing  to  1.8.  The  NLoS  path-loss  exponent   magnitude at higher frequencies; they are  caused
          behaves very differently. At lower frequencies,  as   by  non-obstructed  reflections.  The  NLoS  delay
          the  transmission  losses  are  weak,  the  obstructed   spread tends to slowly decrease versus frequency
          direct  path  often  remains  the  dominant  path,  the   when using a directive antenna (approaching 0 ns
          NLoS n value is quite similar to the ones observed   with the 6° beam). However the NLoS delay spread
          in  the  LoS  situation.  But  the  multi-path  effect   with isotropic antenna has a different behavior. It
          becomes    dominant    at   higher   frequencies,    is  growing  from  2  to  100  GHz, as  the  direct  path
          especially above 28 GHz. As the main component in    suffers from higher attenuation, but some delayed
          the  received  power  comes  from  multiple  non-    indirect paths remain strong.
          obstructed reflected paths, n is rapidly decreasing,
          and  finally  reaches  values  below  1.5  at  150  GHz.
          But this is accompanied by a strong increase in the
          path-loss  intersect  (PLI),  as  shown  in  Fig.  5.  The
          average  difference  between  NLoS  PLI  at  60  GHz
          and 150 GHz is 16 dB while it is 8 dB in LoS. The
          remaining shadowing term S is characterized by its
          standard deviation, as plotted in Fig. 6. It is below
          1 dB in the LoS situation, but rapidly increase with
          frequency  in  NLoS,  in  a  quasi-linear  way  at  the
          highest frequencies.
          A simple sub-THz in-office path-loss model can be
          implemented  as  follows:  the  LoS  probability  is
          given  as  a  function  of  distance;  the  median  path
          loss is calculated from frequency-dependent n and            Fig. 7 – CDF of the delay spread at 150 GHz
          PLI  parameters;  and  the  additional  shadowing
          term S is considered as a log-normal variable with
          a    frequency-dependent     standard-deviation.
          The graphs given in Figures 4-6 are used to derive
          the approximate formulae shown in Table 1, valid
          in the range 90-200 GHz.
             Table 1 – In-office path-loss simplified model (d is the
               distance in meters, and f is the frequency in GHz)








          The effect of  frequency and antenna  directivity is
          also  observed  on  the  delay  spread,  considering  a           Fig. 8 – Median delay spread
          30  dB  power  range  in  the  channel  response     A delay spread that increases with frequency does
          (weakest  simulated  paths  are  filtered  out).  The   not  match  with  common  in-field  observations.
          statistical  distribution  of  the  delay  spread  at   There are two main reasons. Firstly, the obtained
          150 GHz  is  plotted in  Fig.  7.  In  NLoS,  the  median   results are specific to the semi-open confined area
          isotropic  delay  spread  is  divided  by  5  when   that has been simulated; secondly, measurements
          applying a 20° beam,  and divided by  19 with the    are generally affected by a factor that has not been




                                                © International Telecommunication Union, 2019                 21
   32   33   34   35   36   37   38   39   40   41   42