Multi-band Measurements in Indoor Environments

Abstract

This paper presents results of path loss measurements, in two indoor environments for both LoS and NLoS scenarios across a frequency range from 0.6-73 GHz using the multi-band custom designed channel sounders developed at Durham University. The data are analysed to estimate the path loss parameters for each frequency band with either the close in path loss model which assumes free space loss at 1 m reference distance and the floating intercept path loss model which estimates both parameters of the path loss model. The data across the multiple bands are then combined to generate a single set of coefficients for a frequency dependent path loss model. The median and 90% values of the rms delay spread values for a 20 dB threshold are presented across the frequency range of 0.6-39 GHz.

Full text

Multi-band Measurements in Indoor Environments
Sana Salous, Amar Al-Jzari, Mohamed Abdulali, and Jack Towers
Centre for Communication System, Department of Engineering, Durham University, Durham, UK, DH1 3LE
Sana.salous@durham.ac.uk
Abstract—
This paper presents results of path loss
measurements, in two indoor environments for both LoS and
NLoS scenarios across a frequency range from 0.6-73 GHz using
the multi-band custom designed channel sounders developed at
Durham University. The data are analysed to estimate the path
loss parameters for each frequency band with either the close in
path loss model which assumes free space loss at 1 m reference
distance and the floating intercept path loss model which
estimates both parameters of the path loss model. The data
across the multiple bands are then combined to generate a single
set of coefficients for a frequency dependent path loss model.
The median and 90% values of the rms delay spread values for
a 20 dB threshold are presented across the frequency range of
0.6-39 GHz.
Index Terms—
path Loss, office environment, corridor
enviornment, rms delay spread, ITU, millimeter wave
I.
I
NTRODUCTION
Path loss models are necessary for the planning of wireless
networks for the assessment of coverage and interference.
Following the world radiocommunications conference in
2015, several measurements in different scenarios have been
reported in the millimeter wave band for 5G networks. These
were either used to update current international
recommendations such as ITU-R 1411-10 [1] for outdoor
networks or to generate new recommendations such as
building entry loss for outdoor to indoor scenarios, ITU-R
2109-1, [2] and clutter loss for outdoor scenarios, ITU-R
2108-0 [3]. The measurements were mostly classified as line
of sight, LoS or non-line of sight (NLOS). Parameters of path
loss for indoor models in ITU-R 1238-10 were also updated
for several frequencies [4].
Following the adoption of the frequency dependent path
loss model in ITU-R 1411-10, Correspondence Group CG-
3K-6 of working party 3K of Study Group 3 of the ITU,
proposed working towards a similar path loss model in ITU-
R 1238. Currently, the path loss model in ITU-R 1238-10
adopts the close in model with a path loss distance coefficient
for each measured frequency. Working towards updating the
path loss model in ITU-R 1238, measurements across a wide
range of frequencies in indoor scenarios have been conducted
at Durham University. In this paper results of measurements
in a corridor and office scenarios are reported with the
corresponding estimated parameters for the different path
loss models. The rms delay spread values for a 20 dB
threshold are also presented for the median and 90% values.
The rest of the paper is structured as follows. Sections II-
IV describe the measurements and the methodology used
with data analysis in section V, and conclusions in section VI.
II.
MEASUREMENTS METHODOLOGY
The measurements were performed using two chirp
(FMCW) channel sounders developed at Durham University.
One sounder covers three frequency bands, (0.25-1 GHz, 2.2-
2.9 GHz, and 4.4-5.9 GHz). The second sounder has
additional upconverters to an intermediate frequency, IF,
between 12.5-18 GHz with a maximum of 1.5 GHz
bandwidth [5]. The IF signal is then up converted using
frequency multipliers: x2 to generate a 3 GHz bandwidth in
the frequency range 25-29 GHz, x3 to generate 4.5 GHz in
the 36-41 GHz band, x4 to generate 6 GHz in the 50-75 GHz
band and x6 to generate a maximum of 9 GHz in the 60-90
GHz band.
For the current measurements, the two sounders were used
to measure nine frequency bands with a waveform duration
of 819.2 μs, which gives 1.2 kHz waveform repetition rate.
Table 1 gives a summary of the measured frequency bands,
the measured bandwidth, and the corresponding processing
bandwidth. For all the measurements, omni-directional
antennas were used at the transmitter and at the receiver with
the transmit antenna placed at a height close to the ceiling at
2.2-2.6 m as would be deployed in an access point and the
receive antenna at 1.5-1.6 m which is typical for users’
applications.
The measurements were collected over 1-2 seconds and
then processed with bandwidths ranging from 250 MHz to 2
GHz as given in Table 1, to obtain average power delay
profiles (PDP). Each PDP was then used to estimate the
received power above the estimated noise floor similar, to the
procedure followed in [6]. Only PDPs, which met a minimum
of 10 dB SNR, were used in the path loss model. The rms
delay spread was also estimated for 20 dB threshold from the
maximum.
TABLE 1: MEASUREMENTS PARAMETERS, GHz
Freq. 0.63 2.4 4.8 15.6,
17.6
26.8 38.3 62.6 70.3
RF
BW
0.75 0.25 0.5 1.5
3 4.5 6
Analy
-sis
BW
0.25
1
1.5 2
III.
MEASUREMENTS SCENARIOS
Fig. 1 shows two of the measured environments: corridor
and office. In both environments, measurements were
performed in LoS and NLoS scenarios as illustrated in the

layout of Fig. 2 which indicates the positions of the
transmitter antenna and the path followed by the receiver.
(a) (b)
Fig, 1 Measurement scenarios (a) corridor, (b) office
(a)
(b)
(c)
Fig. 2 Layout of measurements (a) corridor, (b) office NLoS, (c) office LoS
IV.
MEASUREMENTS ANALYSIS
Fig. 3 shows the PDPs normalised with respect to the
strongest received component measured in the corridor
environment for both the LoS and NLoS scenarios at 38 GHz
versus the 3D distance which was estimated by taking into
account the height of the transmit and receive antennas.
Following calibration, the path loss coefficients were
estimated using three models: the close in, CI, model as in
ITU-R 1238 where the path loss is given by Eqn. 1 where d
o
is 1 m and L(do) is the corresponding free space loss and n is
the number of floors which is set to zero in the current
analysis as all the measurements were conducted on the same
floor. Eqn. 2 gives the floating intercept (FI) model where
both the distance coefficient α and the intercept coefficient β
are estimated from the measurements. In both the FI and CI
models, path loss coefficients are given for each individual
frequency. The alternative is the ABG, frequency dependent
model which gives a single set of coefficients which can be
applied across a range of frequencies as given in Eqn. 3 where
in addition to the FI coefficients, a third coefficient is
estimated which gives the frequency dependence parameter,
γ as adopted in ITU-R 1411-10. In all three models,
deviations are modelled as a Gaussian distribution with zero
mean and variance equal to σ.




log

/



 
(1)


10log

    (2)
,  10 log

   10γ log

 
(3)

(a)
(b)
Fig. 3 PDP vs location for the corridor environment at 39 GHz (a) LoS, (b)
NLoS
Fig. 4 displays the estimated path loss versus distance for
the LoS scenario in the office environment with the ABG
model. Table 2 gives a summary of the coefficients of the CI
and FI models for the two environments for LoS and NLoS.
Fig. 4. Path loss in the LoS scenario in the office environment
The tables indicate that generally the value of σ is smaller
for the FI model than the CI model which indicates a better
fit. The ABG model parameters are summarized in Table 3
for the two environments. For the LoS case, the frequency
and distance parameters are close to the free space value of 2.
For the NLoS case, the corridor distance coefficient exhibited
a higher value than the office environment due to the wall
blockage leading to higher losses.
TABLE 2: PATH LOSS PARAMETERS FOR THE CI AND FI
MODELS FOR LOS AND NLOS (A) CORRIDOR, (B) OFFICE
(A)
Freq. GHz Corridor
CI/LoS
(α, σ)
CI/NLoS
(α, σ)
FI/LoS
(α, β, σ)
FI/NLoS
(α, β, σ)
0.625 1.65,
2.23
2.31,
2.99
2.06, 23.9,
2.05
3.41, 16,
2.01
2.405 1.69,
2.34
2.63,
1.77
2.14, 35,
2.12 2.6, 40.5, 1.79
4.81 1.57,
2.69
2.78,
1.53
2.25, 38.5,
2.26
2.46, 49.6,
1.38
15.65 1.41,
2.46
2.85,
1.36
1.52, 55.1,
2.47 3.1, 53.5, 1.26
17.57 1.49,
2.22
3.07,
1.55
1.61, 55.9,
2.22
2.82, 60.1,
1.48
26.82 1.58,
2.36
3.21,
1.76
2.15, 54.6,
2.1
3.07, 62.6,
1.75
38.31 1.54,
2.13
3.37,
1.57
1.68, 62.5,
2.12
3.24, 65.6,
1.56
62.6 1.42,
2.27
3.67,
1.7
1.88, 63.2,
2.02
4.09, 63.9,
1.53
70.28 1.66,
3.05
4.15,
1.97
2.49, 60,
2.59
4.79, 63.2,
1.78
(B)
Freq. GHz Office
CI/LoS
(α, σ)
CI/NLoS
(α, σ)
FI/LoS
(α, β, σ))
FI/NLoS
(α, β, σ)
0.625 1.84,
2.69
1.94,
2.91
2.33, 23.4,
2.49
3.11, 15.6,
2.58
2.405 1.77,
3.43
2.24,
2.02
2.33, 34.4,
3.28
2.32, 39.1,
2.03
4.81 1.77,
2.02
2.32,
1.58
2.42, 39.5,
1.53
2.76, 41.3,
1.49
15.65 1.37,
2.07
2.2,
2.53
1.26, 57.4,
2.07
0.938, 69.9,
2
17.57 1.62,
2.22
2.55,
2.24
2.36, 49.7,
1.82 2.03, 63, 2.17
26.82 1.73,
2.49
2.64,
3.14
2.15, 56.8,
2.38
2.24, 65.4,
3.12
38.31 1.59,
2.6
2.82,
3.45
1.41, 65.9,
2.59 1.46, 79, 3.08
62.6 1.82,
2.7
3.76,
4.18 2.35, 63, 2.4 2.58, 81.1,
3.96
70.28 1.95,
3.2
4.06,
3.15
2.52, 63.6,
3.1
3.12, 79.5,
3
TABLE 3: PATH LOSS PARAMETERS FOR THE ABG MODEL
Distance,
m
Environment Los/N
Los
α β
γ
σ
3.8-24 Corridor LoS 2.02 27.9 1.93 2.56
4.6-24 Corridor NLoS 3 25 2.73 2.64
9.3-30.9 Office LoS 2.03 29.5 1.94 3.18
10.9-31.8 Office NLoS 2.13 28.4 2.91 5.3

V. R
ESULTS OF DELAY SPREAD
The data were also analyzed to estimate the rms delay spread
for a 20 dB threshold. Fig. 5 displays the CDF of the rms
delay spread for the NLoS scenario in the office environment
and the corridor with a summary of the median and 90%
values given in Table 4. The results for the 62 and 70 GHz
are not included due to a limited number of data point that
satisfied the 20 dB threshold used in the estimation of the rms
delay spread. Comparing the CDFs in Fig. 5.a shows two
groups of CDF: one group from 0.625 to 4.81 GHz and a
second group for the frequencies from 15.65 to 38.3 GHz.
Figure 5.b shows a different trend versus frequency, where
the frequency bands from 0.625 to 18 GHz, exhibit one group
of CDFs with larger values than the CDFs for the 26 and 38
GHz bands. The frequency band at 38 GHz displays the
lowest values. The bandwidth of analysis for the bands
between 15-26 GHz is equal to 1 GHz whereas the 2 and 5
GHz bands have a smaller analysis bandwidth. While the
analysis bandwidth impacts the estimated delay spread,
overall Fig. 5 indicates that the analysis bandwidth has not
significantly impacted the delay spread values.
VI. C
ONCLUSIONS
In this work, indoor measurements were performed in two
environments: a corridor and an office environment for both
LoS and NLoS scenarios across a frequency range from
0.625-73 GHz. The data were anslysed to estimate path loss
parameters for the CI, FI and ABG models. The results
indicate that the FI results in a lower standard deviation than
the CI model whereas the ABG model provides a convenient
frequency dependent model which can be used across a wide
range of frequencies similar to the model adopted in ITU-R
1411-10. The data were also analysed to estimate the rms
delay spread. The frequency dependence of the rms delay
values varies depending on the environment and the scenario
and does not seem to be highly impacted by the analysis
bandwidth.
(a)
(b)
Fig. 5. CDF of rms delay spread for the NLoS scenario in the (a) corridor
environment, (b) office environmenta
TABLE 4: RMS DELAY SPREAD VALUES
Freq.
GHz Corridor Office
LoS
50%, 90%,
NLoS
50%, 90%
LoS
50%, 90%
NLoS
50%, 90%,
0.625 13.0, 17.8 17.0, 28.3 16.3, 20.7 15.4, 22
2.405 7.9, 12.5 19.9, 27.4 14.4, 21.1 16.7, 21.4
4.81 10.2, 16.8 18.2, 24.1 15.4, 19.8 17.7, 22.2
15.65 12.9, 23.7 11.4, 16.3 17.6, 25.5 17.4, 24
17.57 11.9, 19.8 11.8, 18.3 18.5, 24.9 18.2, 23.2
26.82 11.6, 16.9 12.0, 17.4 15.3, 20.9 12.6, 17.7
38.31 13.2, 23.1 13.2, 19.8 9.7, 17.1 8.2, 16.7
Acknowledgment
The sounder was developed under EPSRC project
PATRICIAN EP/I00923X/1, and its frequency range
extended under funding from EPSRC Impact Acceleration
Account, Ofcom, UK, and Intel, USA. The authors also
acknowledge EU funding under H2020 Marie Schodolowska
Curie project grant 766231 WAVECOMBE- ITN.
R
EFERENCES
[1] ITU-R., Recommendation ITU-R P.1411-10, Propagation data and
prediction methods for the planning of short-range outdoor
radiocommunication systems and radio local area networks in the
frequency range 300 MHz to 100 GHz
[2] ITU-R, "Recommendation ITU-R P.2109-1: Predication of building
entry loss".
[3] ITU-R, "Recommendation ITU-R P.2108-0: Predication of clutter
loss"
[4] ITU-R, Recommendation ITU-R P.1238-10, Propagation data and
prediction methods for the planning of indoor radiocommunication
systems and radio local area networks in the frequency range 300
MHz to 450 GHz
[5] S. Salous, S. Feeney, X. Raimundo and A. Cheema, "Wideband
MIMO Channel Sounder for Radio Measurements in the 60 GHz
Band", IEEE Transactions on Wireless Communic ations, vol. 15, no.
4, pp. 2825-2832, 2016.
[6] Salous, S., Lee, J., Kim, M.D., Sasaki, M., Yamada, W., Raimundo,
X. & Cheema, A.A. (2020). Radio propagation measurements and
modeling for standardization of the site general path loss model in
International Telecommunications Union recommendations for 5G
wireless networks. Radio Science 55(1): e2019RS006924.