This paper presents 28 GHz path loss model based on cluster obtained from channel measurement campaigns with rotating platforms and directional antennas in the indoor and outdoor environments. The transmitter (TX) and receiver (RX) both sweep a large range of angles in the azimuth and elevation plane on account of covering main propagation paths and measuring burden. As the sequence number of cluster increases, the path loss exponents (PLEs) increase while shadow factors also have a growing tendency. The PLE of allclusters is the least because of multi propagation paths, and the LOS PLEs of corridor scenario are less compared with that of office scenario because of corridor's long and narrow structure. This improved model not only considers cluster characteristics, but also unites directional and omnidirectional models into the same framework, which to some extent improves the 5G mmWave channel model.
This paper was supported by National Basic Research Program of China (973) (Grant No. 2012CB316002), National Hightech R&D Program of China (863) (Grant No. 2015AA01A701), National Natural Science Foundation of China (Grant No. 61201192), Science Foundation for Creative Research Group of NSFC (Grant No. 61321061), National S&T Major Project (Grant No. 2017ZX03001011), MOST “Hongkong, Macau and Taiwan" Science Collaboration Project (Grant No. 2014AA01A707), TsinghuaQualcomm Joint Project.
Figure 1
(Color online) Measurement system.
Figure 2
(Color online) Office scenario (4408 room of FIT building, $15.765$ m $\times$ $16.189$ m $\times$ $2.789$ m), there were 4 TX locations and 12 RX locations for a total of 12 TXRX combination locations. In the NLOS situations, the LOS path was obstructed by the concrete pillars in the center of the room. There were objects such as concrete walls and pillars, glass walls and doors, wood doors and cupboard, plastic chairs and desks in the standard office scenario.
Figure 3
(Color online) Corridor scenario (the fourth floor of FIT building, $105$ m $\times$ $2.367$ m $\times$ $2.485$ m), there were 1 TX location and 10 RX locations for a total of 10 TXRX combination locations. There were objects such as concrete walls, glass windows and wood doors in the standard corridor scenario.
Figure 4
(Color online) Outdoor scenario (courtyard beside the FIT building, $50$ m $\times$ $80$ m), there were 1 TX location and 9 RX locations for a total of 9 TXRX combination locations. The TX was fixed outside the window on the second floor of FIT building, and the orientation of horn antenna was $6^{\circ}$ and $19.7^{\circ}$ in the azimuth and elevation plane, respectively. There were objects such as concrete walls, glass windows, plastic chairs and desks, small trees and udershrub in the standard courtyard scenario.
Figure 5
(Color online) Measuring route (the receiver moved along the route in the middle of the courtyard). 9 measuring positions were nearly uniformly distributed in this route whose distance was about 30 m. The sweep range of RX antenna was [$180^{\circ}$,~$180^{\circ}$] and [$20^{\circ}$,~$20^{\circ}$] in the azimuth and elevation plane, respectively.
Figure 6
(Color online) Clustering the MPCs of 11th TXRX combination location in the office (the clusters are sorted by power and cluster 1 is the strongest one among all of the clusters).
Figure 7
(Color online) 28 GHz path loss model (LOS) based on cluster in the office indoor environment. (a) CI model; (b) FI model.
Figure 8
(Color online) 28 GHz path loss model (NLOS) based on cluster in the office indoor environment. (a) CI model; (b) FI model.
Figure 9
(Color online) 28 GHz path loss model (NLOS) based on cluster in the corridor indoor environment. (a) CI model; (b) FI model.
Figure 10
(Color online) 28 GHz path loss model (LOS) based on cluster in the courtyard outdoor environment. (a) CI model; (b) FI model.
Initialize the set of MPCs:$\Omega_n=\left\{\text{MPC}_n(\alpha_n, \theta_{r,n}, \phi_{r,n}, \theta_{t,n}, \phi_{t,n}, \tau_n), n=1, \dots, L\right\}$. 
Choose the strongest path as the centroid of $\text{Cluster}_i$: $\text{MPC}_i=\max\left\{\text{MPC}_n(\alpha_n),\text{MPC}_n \in \Omega_n\right\}$, 
Choose the similar paths from group $\Omega_i$ as the element of $\text{Cluster}_i$ based on $\text{MPC}_i$, 

Case1: $\sqrt{(\theta_{r,n}\theta_{r,i})^2+(\phi_{r,n}\phi_{r,i})^2+(\theta_{t,n}\theta_{t,i})^2+(\phi_{t,n}\phi_{t,i})^2}\leqslant\Delta\theta$, 
Case2: $\sqrt{\tau_n\tau_i}\leqslant\Delta\tau$, 
$\left\{\text{MPC}_n \in \Omega_n\right\}\subseteq \text{Cluster}_i$, 

Obtain $\Omega_{i+1}$ after remove the $\text{Cluster}_i$ from $\Omega_{i}$ and calculate the ratio: $e_i=\frac{P_i}{P_1}=\frac{\sum\alpha_n^2,\text{MPC}_n \in \Omega_{i+1}}{\sum\alpha_n^2,\text{MPC}_n \in \Omega_{1}}$. 

go to end, 

next $i$, 

Return: $\left\{\text{MPC}_n(\alpha_n, \theta_{r,n}, \phi_{r,n}, \theta_{t,n}, \phi_{t,n}, \tau_n), \text{MPC}_n \in \text{Cluster}_i, i \leqslant \textrm{MAX} \right\}$. 
Copyright 2020 Science China Press Co., Ltd. 《中国科学》杂志社有限责任公司 版权所有
京ICP备18024590号1