16.03.2015 Views

"WINNER II Channel Models", ver 1.1, Sept

"WINNER II Channel Models", ver 1.1, Sept

"WINNER II Channel Models", ver 1.1, Sept

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

<strong>WINNER</strong> <strong>II</strong> D<strong>1.1</strong>.2 V<strong>1.1</strong><br />

( τ ' − min( τ '))<br />

τ = sort<br />

. (4.2)<br />

n<br />

n<br />

In the case of LOS condition additional scaling of delays is required to compensate the effect of LOS<br />

peak addition to the delay spread. Heuristically determined Ricean K-factor dependent scaling<br />

constant is<br />

D +<br />

n<br />

2<br />

3<br />

= 0.7705−<br />

0.0433K<br />

+ 0.0002K<br />

0.000017K<br />

, (4.3)<br />

where K [dB] is the Ricean K-factor defined in Table 4-5. Scaled delays are<br />

LOS<br />

τ = τ D , (4.4)<br />

n n<br />

/<br />

they are not to be used in cluster power generation.<br />

Step 6: Generate the cluster powers P.<br />

The cluster powers are calculated assuming a single slope exponential power delay profile. Power<br />

assignment depends on the delay distribution defined in Table 4-5. With exponential delay distribution<br />

the cluster powers are determined by<br />

'<br />

⎛<br />

Pn<br />

= exp<br />

⎜−<br />

⎝<br />

n<br />

r 1⎞<br />

τ<br />

−<br />

10<br />

r<br />

⎟ ⋅<br />

τστ<br />

⎠<br />

and with uniform delay distribution they are determined by<br />

'<br />

P n<br />

−Ζ<br />

10<br />

n<br />

τ (4.5)<br />

⎛ −τ<br />

⎞<br />

n<br />

= exp<br />

⎜<br />

⎟ ⋅ 10<br />

⎝ σ τ ⎠<br />

−Ζ<br />

10<br />

n<br />

, (4.6)<br />

where Ζ n ~ N(0, ζ ) is the per cluster shadowing term in [dB]. A<strong>ver</strong>age the power so that sum power of<br />

all clusters is equal to one<br />

P<br />

'<br />

n<br />

n<br />

=<br />

N<br />

∑ n = 1<br />

P<br />

Assign the power of each ray within a cluster as P n / M, where M is the number of rays per cluster.<br />

Step 7: Generate the azimuth arrival angles ϕ and azimuth departure angles φ.<br />

If the composite PAS of all clusters is modelled as wrapped Gaussian (see Table 4-5) the AoA are<br />

determined by applying in<strong>ver</strong>se Gaussian function with input parameters P n and RMS angle spread σ ϕ<br />

On equation above 1. 4<br />

AoA<br />

2 ln<br />

' = σ AoA<br />

−<br />

C<br />

P<br />

'<br />

n<br />

( P max( P ))<br />

(4.7)<br />

n<br />

n<br />

ϕ<br />

n<br />

. (4.8)<br />

σ = σ is the standard deviation of arrival angles (factor 1.4 is the ratio<br />

ϕ<br />

of Gaussian std and corresponding ”RMS spread”). Constant C is a scaling factor related to total<br />

number of clusters and is given in the table below:<br />

# clusters 4 5 8 10 11 12 14 15 16 20<br />

C 0.779 0.860 1.018 1.090 <strong>1.1</strong>23 <strong>1.1</strong>46 <strong>1.1</strong>90 1.211 1.226 1.289<br />

In the LOS case constant C is dependent also on Ricean K-factor. Constant C in eq. (4.10) is<br />

substituted by C LOS . Additional scaling of angles is required to compensate the effect of LOS peak<br />

addition to the angle spread. Heuristically determined Ricean K-factor dependent scaling constant is<br />

2<br />

3<br />

( <strong>1.1</strong>035<br />

− 0.028K<br />

− 0.002K<br />

0.0001K<br />

)<br />

C LOS C ⋅<br />

+<br />

= , (4.9)<br />

where K [dB] is the Ricean K-factor defined in Table 4-5.<br />

Assign a positive or negative sign to the angles by multiplying with a random variable X n with<br />

uniform distribution to discrete set of {1,–1}, add component N( 0, 5)<br />

random variation<br />

n<br />

n<br />

n<br />

n<br />

LOS<br />

n<br />

~<br />

AoA<br />

Y σ to introduce<br />

ϕ = X ϕ ' + Y + ϕ , (4.10)<br />

where ϕ LOS is the LOS direction defined in the network layout description Step1.c.<br />

Page 39 (82)

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!