Line of sight (LoS)/non-line of sight (NLoS) point identification in wireless communication networks using artificial intelligence

Patent metadata

Field	Value
Publication number	US-12082147-B2
Application number	US-202117460022-A
Country	US
Kind code	B2
Filing date	Aug 27, 2021
Priority date	Sep 18, 2020
Publication date	Sep 3, 2024
Grant date	Sep 3, 2024

Abstract

A method for performing line-of-sight (LoS)/non-line-of-sight (NLoS) filtering includes collecting, from at least one user equipment (UE) operating within a coverage area of a base station, a plurality of measurement reports from the at least one UE, wherein each measurement report of the plurality of measurement reports comprises a measured value of a signal parameter and location information of the at least one UE. The method further includes training a LoS/NLoS classification model on the plurality of measurement reports, obtaining a new measurement report from a UE operating at a first location within the coverage area of the base station and passing the new measurement report from the UE to the trained LoS/NLoS classification model to obtain a LoS/NLoS classification of the first location.

First claim

What is claimed is:

1. A method for performing line-of-sight (LoS)/non-line-of-sight (NLoS) filtering, comprising:
collecting, by a base station, from at least one user equipment (UE) operating within a coverage area of the base station, a plurality of measurement reports from the at least one UE, wherein each measurement report of the plurality of measurement reports comprises a measured value of a signal parameter and location information of the at least one UE;
selecting, by the base station, a random forest LoS/NLOS classification model from among multiple LoS/NLOS classification models based on a topography of the coverage area of the base station;
training, by the base station, the selected random forest LoS/NLOS classification model on the plurality of measurement reports by generating multiple decision trees based on randomly selected subsets of a set of input parameters and polling across the decision trees to identify predictive features, wherein the randomly selected subsets of input parameters correspond to parameters contained in the measurement reports;
obtaining, by the base station, a new measurement report from a UE operating at a first location within the coverage area of the base station; and
passing, by the base station, the new measurement report from the UE to the trained random forest LoS/NLOS classification model to obtain a LoS/NLOS classification of the first location of the UE.

2. The method of claim 1 , further comprising:
collecting, from the at least one UE operating within the coverage area of the base station, a second plurality of measurement reports from the at least one UE, wherein measurement reports of the second plurality of measurement reports comprise measurements taken at LoS locations and NLOS locations within the coverage area of the base station;
passing the second plurality of measurement reports to the trained random forest LoS/NLOS classification model to identify measurement reports from LoS locations within the coverage area of the base station; and
tuning a transmission parameter of the base station based on the identified measurement reports from LoS locations within the coverage area of the base station.

3. The method of claim 2 , wherein the transmission parameter of the base station comprises at least one of an antenna height, an antenna pattern, an antenna azimuth angle, a mechanical tilt (“M-Tilt”) angle or an electrical tilt (“E-Tilt”) angle of the base station.

4. The method of claim 1 , wherein the LoS/NLOS classification of the first location is at least one of a binary classification of the first location as LoS or NLOS, or a probabilistic representation of a relative likelihood of the first location being LoS or NLoS.

5. The method of claim 1 , further comprising:
transmitting, by the base station, to the at least one UE, a trigger message for collecting measurement report data and transmitting a measurement report to the base station; and
transmitting by the base station, to the at least one UE, a reference signal for measurement by the at least one UE.

6. The method of claim 1 , wherein the measured value of the signal parameter comprises at least one of a value of a reference signal received power (RSRP) from the base station, a value of an RSRP from a neighbor base station, or a reference signal received quality (RSRQ) from the base station.

7. A base station comprising:
a processor;
a network interface; and
a memory containing instructions, which when executed by the processor, cause the base station to:
collect, via the network interface, a plurality of measurement reports from at least one UE operating within a coverage area of the base station, wherein each measurement report of the plurality of measurement reports comprises a measured value of a signal parameter and location information of the at least one UE,
select a random forest LoS/NLOS classification model from among multiple LoS/NLOS classification models based on a topography of the coverage area of the base station,
train the selected random forest LoS/NLoS classification model on the plurality of measurement reports by generating multiple decision trees based on randomly selected subsets of a set of input parameters and polling across the decision trees to identify predictive features, wherein the randomly selected subsets of input parameters correspond to parameters contained in the measurement reports,
obtain, via the network interface, a new measurement report from a UE operating at a first location within the coverage area of the base station, and
pass the new measurement report from the UE to the trained random forest LoS/NLOS classification model to obtain a LoS/NLOS classification of the first location of the UE.

8. The base station of claim 7 , wherein the memory
contains instructions, which when executed by the processor, cause the base station to:
collect, via the network interface, from the at least one UE operating within the coverage area of the base station, a second plurality of measurement reports from the at least one UE, wherein measurement reports of the second plurality of measurement reports comprise measurements taken at LoS locations and NLOS locations within the coverage area of the base station,
pass the second plurality of measurement reports to the trained random forest LoS/NLOS classification model to identify measurement reports from LoS locations within the coverage area of the base station, and
tune a transmission parameter of the base station based on the identified measurement reports from LoS locations within the coverage area of the base station.

9. The base station of claim 8 , wherein the transmission parameter of the base station comprises at least one of an antenna height, an antenna pattern, an antenna azimuth angle, a mechanical tilt (“M-Tilt”) angle or an electrical tilt (“E-Tilt”) angle of the base station.

10. The base station of claim 7 , wherein the LoS/NLOS classification of the first location is at least one of a binary classification of the first location as LoS or NLoS, or a probabilistic representation of a relative likelihood of the first location being LoS or NLoS.

11. The base station of claim 7 , wherein the memory further contains instructions, which when executed by the processor, cause the base station to:
control the base station to transmit to the at least one UE, a trigger message for collecting measurement report data and transmitting a measurement report to the base station, and
control the base station to transmit to the at least one UE, a reference signal for measurement by the at least one UE.

12. The base station of claim 7 , wherein the measured value of the signal parameter comprises at least one of a value of a reference signal received power (RSRP) from the base station, a value of an RSRP from a neighbor base station, or a reference signal received quality (RSRQ) from the base station.

13. A non-transitory computer readable medium containing instructions that when executed cause at least one processor of a base station to:
collect a plurality of measurement reports from at least one UE operating within a coverage area of the base station, wherein each measurement report of the plurality of measurement reports comprises a measured value of a signal parameter and location information of the at least one UE;
select a random forest LoS/NLOS classification model from among multiple LoS/NLOS classification models based on a topography of the coverage area of the base station;
train the selected random forest LoS/NLOS classification model on the plurality of measurement reports by generating multiple decision trees based on randomly sele

Assignees

• Samsung Electronics Co Ltd

Inventors

• Shafin Rubayet
• RATNAM Vishnu Vardhan
• Chen Hao
• Xin Yan
• Zhang Jianzhong

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/0895

  Weakly supervised learning, e.g. semi-supervised or self-supervised learning · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• H04W24/10 Primary

  Scheduling measurement reports {; Arrangements for measurement reports} · CPC title

• G06N20/00

  Machine learning · CPC title