Page 191 - Kaleidoscope Academic Conference Proceedings 2021

Connecting physical and virtual worlds

function. The optimizer adopts the stochastic gradient It can be seen that this ratio is relatively large for the LOS
descent algorithm. channel scenario, and is relatively small for the NLOS
channel scenario, which can basically distinguish
4. SIMULATION LOS/NLOS scenarios.

Table 2 gives the simulation parameters for wireless 1 CDF of frequency domain fading factor
channel scenario recognition. 3.5G Hz is the global main 0.9 LOS,Low-Speed
LOS,Mid-Speed
stream frequency band, and the 30k Hz subcarrier spacing 0.8 LOS,High-Speed
NLOS,Low-DS,Low-Speed
is also a common configuration in 5G systems. In order to 0.7 NLOS,Low-DS,Mid-Speed
NLOS,Low-DS,High-Speed
calculate the time correlation characteristics of the pilot 0.6 NLOS,High-DS,Low-Speed
NLOS,High-DS,Mid-Speed
signal, this paper chooses the configuration of two pilot Probabilty 0.5 NLOS,High-DS,High-Speed
0.4
symbols. Since the channel characteristics of the spatial 0.3
dimension are not involved in this paper, the number of 0.2
physical antennas of the base station and the UE are 2 and 1 0.1
respectively. Due to space limitations, this section only 0 0 0.5 1 1.5 2 2.5 3 3.5
presents the distribution curve of wireless channel normalized value
characteristics and the performance of the two classification
algorithms when the UE occupies 50 resource blocks. Figure 3 – CDF of frequency domain fading factor ( )
Table 2 – Simulation parameters

Parameter Value
Carrier frequency 3.5GHz
Subcarrier spacing 30kHz
Sample rate 1/(4096×30e3)
Antenna number of base station 2
Antenna number of user 1
equipment (UE)
Allocated resource block of UE 50
Pilot signal overhead 2 symbols Figure 4 – Multipath power delay distribution ( , )in the
time domain
The simulation models nine wireless channel scenarios
proposed in Table 1, respectively using 4G and 5G channel CDF of Power peak response ratio in time domain
models. These channel models are defined by a 3rd 1 LOS,Low-Speed
Generation Partnership Project (3GPP), such as the 0.9 LOS,Mid-Speed
LOS,High-Speed
NLOS,Low-DS,Low-Speed
0.8
Extended Pedestrian A model (EPA), Extended Vehicular 0.7 NLOS,Low-DS,Mid-Speed
NLOS,Low-DS,High-Speed
NLOS,High-DS,Low-Speed
A model (EVA) and the Extended Typical Urban model 0.6 NLOS,High-DS,Mid-Speed
NLOS,High-DS,High-Speed
(ETU), Clustered Delay Line (CDL) and Tapped Delay Probabilty 0.5
Line (TDL). Where CDL-A, CDL-B, CDL-C, TDL-A, 0.4
TDL-B and TDL-C are used to simulate NLOS channel 0.3
scenarios, CDL-D, CDL-E, TDL-D and TDL-E are used to 0.2
0.1
simulate the LOS channel scenarios.
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
normalized value
4.1 Distribution curve of channel characteristics
Figure 5 – CDF of channel power peak response ratio ( )
In order to better understand the influence of channel in the time domain
characteristics on the classification of channel scenarios, we
give the distribution curve of each channel characteristic for 1 CDF of channel correlation value in time domain
nine wireless channel scenarios. 0.9 LOS,Low-Speed
LOS,Mid-Speed
LOS,High-Speed
0.8 NLOS,Low-DS,Low-Speed
Figure 3 shows the Cumulative Distribution Function (CDF) 0.7 NLOS,Low-DS,Mid-Speed
NLOS,Low-DS,High-Speed
curve of the frequency domain fading factor. It can be seen 0.6 NLOS,High-DS,Low-Speed
NLOS,High-DS,Mid-Speed
NLOS,High-DS,High-Speed
that when the multipath delay spread increases, the value of Probabilty 0.5
0.4
this factor also increases, which reflects the influence of 0.3
time dispersion on frequency domain selective fading. 0.2
Figure 4 shows the multipath power delay distribution. 0.1
When the multipath delay spread increases, the multipath 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
power distribution becomes more scattered. Figure 5 shows normalizated value
the channel power peak response ratio in the time domain.
Figure 6 – CDF of channel time correlation value ( )
– 129 –