Competitors

Competing Teams 2023


Offsite: 11-15 Sep, EvaalAPI web service
Onsite: 23-24 Sep, Museum for Industrial Culture, Nuremberg (DE)
Awards: 28 Sep, IPIN conference

 

Track 1 - Smartphone (on-site)

 

Team Name

THWS


Team members

Markus Ebner, Steffen Kastner, Markus Bullmann, Toni Fetzer, Frank Deinzer


Affiliation

Technische Hochschule Würzburg-Schweinfurt


Description

Based on a condensation particle filter, the localization system is  highly modular with respect to the sensors it is able to use.
Models exist for Bluetooth low energy beacons, Wi-Fi, and others.
Combined with a step and turn detection, it can also handle scenarios of total sensor outages.
All calculations are performed in real-time on a commercial smartphone  even when using a high number of samples for approximation.


 

Team Name

Inha


Team members

Wonik Choi, Chungheon Yi , Taeyeab Kim, Young-jun Jeon, Minjoon Jeon, Kiryang Kwon


Affiliation

Inha University, Papaya Co., LTD.


Description

PapayaIPS is an artificial intelligence-based positioning solution,  which employs a specially designed comparative neural network to  estimate locations. We develop our own comparative neural network and  call this neural network PNN (Positioning Neural Network). PapayaIPS is  basically aiming to utilize all signals that can be received by a  smartphone such as Wi-Fi, BLE, barometer, and accelerometer, etc., to  provide a comprehensive positioning solution.
PapayaIPS is based on  signal tensor images that are transformed and visualized from those  multi-dimensional signal values received by a smartphone at a certain  point. Then, PNN learns to transform the similarity of two signal  tensors into the Euclidean distance indoors by training with these  signal tensor images.


Team Name

KNU WCSL


Team members

Jeongsik Choi


Affiliation

Kyungpook National University


Description

We consider a well-known positioning approach which is based on Wi-Fi  ranging and PDR method. Such an approach requires to secure the location  database of Wi-Fi access points (APs) and investigate the signal  propagation characteristics under each site for precise ranging. To  minimize human intervention to conduct such a site survey task, we apply  an automatic site survey method that estimates the locations of APs and  learn signal propagation characteristics simultaneously using  measurement data provided from multiple mobile users.


Track 3 - Smartphone (off-site online)

Team Name
IPNL


Team members

Qing Liang


Affiliation

The Hong Kong Polytechnic University


Description

The system follows the particle filter-based multi-sensor fusion  positioning framework using PDR, WiFi RSSI fingerprinting, and map  matching. AHRS heading estimation accuracy is critical to good PDR  performance. Here we use the magnetometer alongside IMU for accurate  attitude and low-drift heading estimation. In particular, quasi-static  magnetic fields are detected to correct gyroscope biases and reduce  heading drift. Opportunistic magnetic headings that are unaffected by  magnetic disturbances are identified to provide occasional absolute  heading correction.
PDR entails step detection, step length  estimation, and heading change estimation. Step events are detected on  the vertical acceleration by examing the periodic human motion patterns.  We use an adaptive zero-crossing detector and finite-state machine for  reliable step detection. We choose the Weinberg step length model as it  is easy to tune with only one parameter.


Team Name
IOT2US


Team members

Yonglei Fan, Qiqi Shu, Zhao Huang, Guangyuan Zhang, Meng Xu, Xijie Xu, Guangxia Yu


Affiliation

Queen Mary, University of London, Peking University


Description

We mainly use the acce, gyro and Mega to estimate the step length, step and direction.
it  is real time indoor positioning method, we will separate all data into  small pieces and use these pieces of data to do the positioning.
For the first positioning, we may use WiFi, or Bluetooth data.
Kalman  filter will be used to minimize the error and the cumulative error. But  there must be some cumulative error which is hard to deal with. We  should think a better way to fix it.
Pedestrian dead reckoning (PDR)  has become a research hotspot since it does not require a positioning  infrastructure. An integral equation is used in PDR positioning; thus,  errors accumulate during long-term operation. To eliminate the  accumulated errors in PDR localisation, we proposes a PDR localisation  system applied to complex scenarios with multiple buildings and large  areas. The system is based on the pedestrian movement behavior  recognition algorithm proposed.                  


Team Name
IA3


Team members

Alex Martinez-Martinez, Javier Gabiña Rueda


Affiliation

Universitat Jaume I


Description
The system aims to achieve accurate  indoor positioning through the integration of various technologies such  as Pedestrian Dead Reckoning (PDR), WiFi fingerprinting, multi-sensor  fusion, activity recognition, and map information. Core components  include multi-sensor fusion, WiFi fingerprinting, and PDR prediction.  These components work together to predict the user's position using  sensors and WiFi data. Activity recognition and map information  complement the predictions. Machine learning models such as LSTMs, RNNs,  and CNNs are employed. The system enhances accuracy by identifying  floor changes and predicting turns. It utilizes map information to guide  users and ensure they stay within building boundaries. The goal is to  provide highly accurate and reliable indoor positioning.


Team Name
QAQ


Team members

Zheng Shiyu


Affiliation

Xiamen University


Description
Our multi-source indoor positioning  system integrates data from an Inertial Measurement Unit (IMU), Wi-Fi,  magnetometer, barometer, and indoor maps to achieve precise indoor  localization.
The system utilizes the Inverted Pendulum Model-based  Pedestrian Dead Reckoning (PDR) technique to estimate the user's  position.
In addition to PDR, the system incorporates Wi-Fi and  Bluetooth data to create fingerprint maps. Fingerprint matching is  employed to determine the user's current position, while changes in  barometric pressure aid in floor level detection. The data from each  component is fused using a factor graph-based approach, specifically  based on gtsam, to achieve accurate and reliable positioning results.


 

Team Name
Inha


Team members

Wonik Choi, Chungheon Yi , Taeyeab Kim, Young-jun Jeon, Minjoon Jeon, Kiryang Kwon


Affiliation

Inha University, Papaya Co., LTD.


Description
PapayaIPS is an artificial  intelligence-based positioning solution, which employs a specially  designed comparative neural network to estimate locations. We develop  our own comparative neural network and call this neural network  PNN(Positioning Neural Network). PapayaIPS is basically aiming to  utilize all signals that can be received by a smartphone such as Wi-Fi,  BLE, barometer, and accelerometer, etc., to provide a comprehensive  positioning solution.
PapayaIPS is based on signal tensor images  that are transformed and visualized from those multi-dimensional signal  values received by a smartphone at a certain point. Then, PNN learns to  transform the similarity of two signal tensors into the Euclidean  distance indoors by training with these signal tensor images.



Team Name
KNU WCSL


Team members

Jeongsik Choi


Affiliation

Kyungpook National University


Description
We consider a well-known positioning  approach which is based on Wi-Fi ranging and PDR method. Such an  approach requires to secure the location database of Wi-Fi access points  (APs) and investigate the signal propagation characteristics under each  site for precise ranging. To minimize human intervention to conduct  such a site survey task, we apply an automatic site survey method that  estimates the locations of APs and learn signal propagation  characteristics simultaneously using measurement data provided from  multiple mobile users.


Team Name
SZU

Team members

Mengyuan Tang


Affiliation

Shenzhen University, College of Civil and Transportation Engineering


Description
Our system uses deep recurrent neural  networks to learn the sensor data from smartphones. Based on sensor data  and BLE beacons information, we train to recognize the basic behaviors  of pedestrians inside buildings (detecting behaviors like going  upstairs, in a lift, turning, etc.), and integrate the attitude results  of pedestrians dead reckoning (PDR) methods to achieve trajectory  positioning of pedestrians in multi-floor buildings.
Nowadays,  Indoor space has been an important space for human activities and people  spend more than 80% of their time in the indoor environment. However,  as a common scenario in urban indoor spaces, multi-floor buildings face  issues such as missing or attenuated GNSS signals. Therefore, it is one  of the current research hotspots to realize the robust low-cost  navigation and positioning in complex indoor environments.
Our system uses the PDR methods based on the Inertial Measurement Unit (IMU) to estimate the three-dimensional position.


Team Name
BUPTer


Team members

Chaoyi Xu, Fan Yin


Affiliation

Beijing University of Post and Telecommunication, China (BUPT)


Description
Our project aims to utilize an  Inertial Measurement Unit (IMU) as the principal navigation sensor,  augmented by additional sensors, to realize precise positioning.  Notably, the inherent integration process of Pedestrian Dead Reckoning  (PDR) might cause cumulative errors, especially when mobile-phone grade  IMUs, which are less accurate and more prone to individual's walking  styles, are employed. Hence, our project proposes to incorporate WiFi  fingerprint data to assist the inertial navigation system, leveraging a  fuzzy matching technique to mitigate the accumulation of errors. We also  plan to implement a filtering and fusion algorithm to reduce  disturbances from electromagnetic interference or sensor errors.
Further, we use a subset of actual location data to intermittently  calibrate for path deviations. It may also involve the preliminary  conversion of longitude and latitude coordinates into a planar  coordinate system.


 

Team Name

THWS


Team members

Markus Ebner, Steffen Kastner, Markus Bullmann, Toni Fetzer, Frank Deinzer


AffiliationIOT2US

Technische Hochschule Würzburg-Schweinfurt


Description
Based on a condensation particle  filter, the localization system is highly modular with respect to the  sensors it is able to use.
Models exist for Bluetooth low energy beacons, Wi-Fi, and others.
Combined with a step and turn detection, it can also handle scenarios of total sensor outages.
All calculations are performed in real-time on a commercial smartphone  even when using a high number of samples for approximation.


 

Team Name

imec-Waves


Team members

Cedric De Cock, David Plets


Affiliation

Ghent University/imec-Waves


Description
Our system consists of six modules, which are briefly described in this document.
The first module is an interface for the Evaal server, which starts the  Evaal trial and requests the next stream of smartphone data in blocks  of 0.5 s, and uploads the estimated position. The second modules is a  Pedestrian Dead reckoning (PDR) algorithm, which detects steps from the  smartphone’s IMU data, and estimates the step length and heading. The  third module is a 3D graph of the environment, in which the nodes are  accessible positions, and the edges are physically valid (for a human)  ways of traveling between the nodes. The fourth module performs BLE  fingerprinting, matching new RSS measurements using KNN. The fifth  modules is a floor (transition) detection algorithm, which (f)uses  barometer, accelerometer, GPS, RSS data. The sixth module is a tracking  algorithm which incorporates the graph, PDR, fingerprinting, and floor  (transition) detection to estimate the user position.


Track 4 - Foot-mounted IMU (off-site online)

 

Team Name

ININ624


Team members

Zhidong Meng


Affiliation

School of Automation, Beijing Institute of Technology


Description
The Calibration of sensors is carried  out leveraging the given data including constant bias estimation,  temperature compensation, and random characteristics for the gyroscope.  The calibration of magnetometer is realized by ellipsoid fitting method.  The initial velocity is set to 0, and the initial position is obtained  by GNSS solution. The initial attitude is determined by TRIAD method.  The error state sequential Kalman filter is used to estimate the errors  in attitude, velocity, position, accelerometer drifts, gyroscope drifts,  and body odometer (BOR) drifts. The observation resources from pseudo  Zero-velocity measurement, Barometer, Global Navigation Satellite  System, and BOR. Fusing BOR is an algorithm proposed by our team that  utilizes displacement estimation to suppress multiple errors in inertial  navigation by breaking the integral process. To avoid failed observer  to disturb measurement updates, a chi square test and expectation  limitation are introduced.


Team Name

CETC-CePNT


Team members

Baoguo Yu, Jun Li, Xinjian Wang, Yanan Hu, Haonan Jia, Lu Huang


Affiliation

The 54th Research Institute of China Electronics Technology Group Corporation


Description
Multiple conditions zero-velocity  detection algorithms are employed to detect zero-velocity intervals  within the walking gait. Within the detected zero-velocity intervals,  the principle of zero-velocity correction algorithm is utilized to  construct the observed quantity of velocity error; leveraging the  characteristics of stationary pedestrian foot-ground contact and only  subjected to gravity acceleration and unchanged posture angles within  the zero-velocity intervals to construct the observed quantity of  posture angle error. Satellite signals acquired by the system are  differentiated and identified, usable signals are extracted, and a  Kalman filter is applied to estimate errors in posture angles, velocity,  and position within the zero-velocity intervals. The obtained error  state estimation results are utilized for error correction in pedestrian  navigation, enhancing the accuracy of inertial pedestrian navigation.


Team Name

VINF


Team members

Jiale Han, Maoran Zhu, Yuanxin Wu


Affiliation

Institute for Sensing and Navigation, Shanghai Jiao Tong University


Description
In our system design, the core is an  information fusion algorithm based on the Error State Kalman Filter with  five constraints, including the Zero-velocity update (ZUPT), the Zero  angular rate update (ZARU), the Improved heuristic drift elimination  (iHDE), the Ellipsoid constraint, and the Constant speed.

Additionally, the height variation calculated by the pressure  sensor can also be used to correct the navigation state. The magnetic  field will be used for loop detection, enabling the utilization of  historical estimated positions to correct the current navigation state.  When the user comes to an outdoor scene, the received GNSS signals will  be used to correct the current navigation state.
The parameters of  inertial sensors, such as the bias instability of gyroscopes and  accelerometers, the angle random walk, and the velocity random walk, are  determined through long-term static data.


Team Name

SmartLoc


Team members

Wang Han, Zhang Mingkai


Affiliation

Nanyang Technological University


Description
Indoor localization has emerged as a  crucial technology for a wide array of applications such as robotics,  asset tracking, and indoor navigation. Traditional GPS-based methods  face limitations when applied indoors due to signal attenuation and  multipath effects. To address these challenges, an innovative indoor  localization method has been developed, leveraging an error-state Kalman  filter that integrates data from Inertial Measurement Units (IMUs),  Global Positioning System (GPS) information, and zero velocity updates.  This comprehensive approach enhances the accuracy and reliability of  indoor positioning, making it ideal for environments where GPS signals  are not available or are unreliable.


Track 5 - Smartphone (off-site online)

Team Name

BUPTer


Team members

Fan Yin, Chaoyi Xu


Affiliation

Beijing University of Post and Telecommunication, China (BUPT)


Description Our project aims to achieve precise  positioning by utilizing an Inertial Measurement Unit (IMU) as the  primary navigation sensor, along with additional sensors such as  Bluetooth Low Energy (BLE) and Received Signal Strength Indicator  (RSSI). However, we acknowledge that the inherent integration process of  Pedestrian Dead Reckoning (PDR) using mobile-phone grade IMUs, which  are less accurate and more prone to individual walking styles, can lead  to cumulative errors. Therefore, we propose to incorporate BLE and RSSI  data to assist the inertial navigation system, leveraging a fuzzy  matching technique to mitigate the accumulation of errors.
To reduce  disturbances from electromagnetic interference or sensor errors, we  plan to implement a filtering and fusion algorithm using only the  available accelerometer and gyroscope data.


Team Name

SmartLoc


Team members

Wang Han


Affiliation

Nanyang Technological University


Description
Basically, in this track the main  challenge is to use the inertial information along with the beacon and  map. The system is designed based on ROS and C. We leverage an extended  Kalman filter and particle filer to achieve the real-time localization.  Then, on the second test, where bluetooth is provided, we form a graph  system. a node if placed every 1 second, with the bluetooth measurement  result. we minimize the displacement error and an bluetooth penalty if  the corresponding beacon is not nearby.


Team Name

UCLab


Team members

Kazuma Kano, Keisuke Higashiura, Kohei Yamaguchi, Koki Takigami, and Yoshiteru Nagata


Affiliation

Nagoya University


Description
We integrated deep-learning-based  Pedestrian Dead reckoning (PDR), map-matching, multilateration, and  fingerprinting on particle-filter. We analyzed heading estimation  reliability and use absolute heading in reliable areas and relative  heading in unreliable areas.


Team Name

Kaji Lab


Team members

Ryuki Toyama


Affiliation

Aichi Institute of Technology


Description

There are from 1 to 6 steps in the process.
1. estimate the  timing of the gait from the acceleration and integrate the angular  velocity to estimate the direction of motion at each time. Based on  this, a two-dimensional gait trajectory is generated. 2.
2. rotate  the entire walking trajectory based on the coordinates of the BLE beacon  that emits strong radio waves before the half of the entire walking  time.
3. perform a grid search for drift based on the final  coordinates of the walking trajectory and the coordinates of the end  point, and generate a walking trajectory with correction for the  obtained drift.
4. rotate the entire trajectory based on the  coordinates of the BLE beacon that emits strong radio waves during the  entire walking time.
5. same as step 3. 6.
6. perform a simple  map-matching correction. If the walking trajectory exists at a point  that does not exist on the map, the trajectory is shifted to a point  that does exist.


Team Name

Dream Hunters


Team members

Fu Han,Naoki Tanabe, Jiang Jiayin, Hiroya Yamashiro


Affiliation

University of Tsukuba, Master's Programs in Intelligent and Mechanical Interaction Systems


Description

This system is based on sample code aimed to estimate the indoor  position of pedestrians by integrating PDR and BLE beacon reception  data. We are planning to make modifications to this system before the  actual competition takes place.
The estimation process utilizes IMU  data (acceleration, gyro, geomagnetism) and BLE beacon reception  information collected from the 9-axis IMU sensor attached to the  pedestrian's AQUOS Sense 6 device (SHARP) within a commercial facility.  Ultimately, accuracy is evaluated through comparison with ground truth  data.


Team Name

CARELab


Team members

Pereira Matthieu, Huakun Liu, Yutaro Hirao, Monica Perusquia-Hernandez,Hideaki Uchiyama, and Kiyoshi Kiyokawa


Affiliation

Nara Institute of Science and Technology


Description

The solution proposed is based on a fingerprinting algorithm. A  fingerprinting algorithm consists of making a map associating the  characteristics of the data sensed with a ground truth position during a  phase of training, then comparing the data sensed with the map during  the testing.


 

Team Name

imec-Waves


Team members

Cedric De Cock, David Plets


Affiliation

Ghent University/imec-Waves


Description
Our system consists of six modules, which are briefly described in this document.
The first module is an interface for the Evaal server, which starts the  Evaal trial, requests and parses the sensor data, then provides the  data to the positioning system. It also uploads the estimated position  sequence the Evaal server. The second module is a PDR algorithm, which  detects steps from the smartphone’s IMU data, and estimates the step  length and heading. The third module is a 3D graph of the environment,  in which the nodes are accessible positions, and the edges are  physically valid (for a human) ways of traveling between the nodes. The  fourth module performs BLE fingerprinting, matching new RSS measurements  using KNN. The fifth module is a floor (transition) detection  algorithm, which (f)uses accelerometer and BLE data. The sixth module is  a tracking algorithm which incorporates the graph, PDR, fingerprinting,  and floor (transition) detection to estimate the user position.


Track 6 - Smartphone on vehicle (off-site online)


Team Name

SmartLoc


Team members

Han Wang


Affiliation

TODO


Description
TODO


Team Name

BJTU-DiDi


Team members

Shuli Zhu, Yufei Su, Feng Liu, Haitao Li, Xuan Xiao, Yuqin Jiang, Ruipeng Gao


Affiliation

School of Software Engineering, Beijing Jiaotong University


Description

Different from the traditional speed estimation method based on  attitude calculation and acceleration integration, we designed a  supervised deep learning model (Speed DNN Model) for estimating vehicle  speed, and estimated the vehicle heading (Bearing Estimation) decoupled  from the vehicle speed estimation function in indoor environments.  Meanwhile, in outdoor environments, we utilize Extended Kalman Filter  (EKF) in order to improve the location of vehicle when GNSS signals and  IMU in the smartphone are valid.

The overall workflow of DNN  model is divided into two lines: model training and model inference.  During model training, we use the provided testing data with  high-quality GNSS information to build a training dataset with  ground-truth speeds. Among them, we have designed two schemes for the  use of inertial data. The first solution is to input the original  sampled data (Raw Data) into the deep learning model without additional  processing of the inertial data.


Team Name

AINS


Team members

Zhuang Guangchen


Affiliation

Beijing Automation Control Equipment Institute


Description

In the first 5 minutes, we use a nonlinear estimation algorithm  based on geomagnetic and gravity vectors to obtain the initial attitud.  During the validity period of GNSS, we estimate and compensate the  installation errors between the IMU, odometer, geomagnetic sensor, and  the vehicle. After the degradation of GNSS occurs, the satellite signals  are evaluated through a model-based detection method. When the accuracy  of satellites cannot meet the requirements of precision, they are  promptly excluded from the integrated navigation framework to reduce the  negative impact to the navigation system. When the satellite signal  fails completely, vehicle navigation and positioning based on fully  autonomous sensors are used. A constraint mechanism based on vehicle  motion is introduced to further improve the robustness of autonomous  navigation and positioning.


Track 7 - 5G CIR (off-site online)

Team Name

WSL Hanayng


Team members

Sunwoo Kim, Paulson Eberechukwu N, Minsoo Jeong, Hongseok Jung, Suah Park


Affiliation

Hanyang University


Description
The WSL team proposes a convolution  neural network (CNN)-based stacked autoencoder approach for indoor  localization that leverages Channel impulse responses (CIR) and  Time-of-arrival (TOA) fingerprints. The proposed approach utilizes a  stacked autoencoder network to mitigate the effects of noise and signal  fluctuations arising from Non-Line-of-Sight (NLOS) scenarios. The  network is trained on pairs of noisy and fluctuating signals and their  corresponding clean versions. By minimizing the reconstruction error,  the network learns to denoise and restore signals affected by NLOS. The  trained stacked autoencoder network effectively reduces noise and  mitigates signal fluctuations, enhancing the accuracy and reliability of  indoor localization systems in challenging NLOS environments.


Team Name

Byr_Trackers


Team members

Zhou Heyang, He Jiawei, Song Xudong, Li Shude, Zhang Haoyu, Dong  Congrong, Zhang Zhichao, Ding Zhenke, Liu Bingxun, Ma Mingyang, Wang  Jizhou


Affiliation

Beijing University of Posts and Telecommunications


Description
The proposed system consists of two  main components: a fingerprint-based positioning system using 5G CSI for  non-line-of-sight (NLOS) scenarios and a TOA based positioning system  for line-of-sight (LOS) scenarios.
In the fingerprint positioning  system, we want use a dual-branch CNN network that consists of two  parallel sub-networks for amplitude and phase information, respectively.  Each sub-network consists of multiple convolutional layers and pooling  layers, followed by fully connected layers. The main algorithm of TOA  based positioning system is the Nonlinear Least Squares positioning  algorithm based on bilateral grid filtering.


Team Name

EURECOM


Team members

Mohsen AHADI


Affiliation

EURECOM


Description

Collect ToA from 8 anchors. K-means cluster ToA for LOS/NLOS. Min  ToA anchor is reference. Calc LOS TDoA. Extract CIR for TDoA anchors.  Sequential NN: input CIR, predict TDoA. Particle Swarm Optimization for  Position estimation at the end.


Team Name

ISCAS


Team members

Chang Su, Fusang Zhang, Beihong jin


Affiliation

Institute of Software, Chinese Academy of Sciences


Description

First, we use machine learning methods that take some hand-crafted  features, e.g., the distance corresponding to the highest peak in the  CI, as the input to estimate the distance between the transmitter and  receiver. Next, we collect the range estimation from anchors with the  same burst-id and use a trilateration algorithm to calculate the  target's position. Finally, we use filters to smooth the trajectory and  resample it to get the final results.


Track 8 - 5G ToF (off-site online)


Team Name

HHULGD


Team members

Huang Ao, Wang Qian, Cui Zhichao, Chen Liang, Cui Yang


Affiliation

Hohai University、Army Engineering University of PLA


Description
AI algorithm estimation localization
We use the AI localization method to establish a mapping scheme between  position and measurement through machine learning algorithms and neural  networks, thus achieving localization.
One of the algorithms is the  ELM algorithm, which is currently a type of deep learning. ELM randomly  selects the weight of the input layer and the bias of the hidden layer,  and the weight of the output layer is calculated analytically based on  the (Moore Penrose, MP) generalized Inverse matrix theory by minimizing  the Loss function composed of the training error term and the regular  term of the weight norm of the output layer. Extreme learning machine  has the advantages of less training parameters, fast learning speed and  strong generalization ability.
In addition, for Track-8, we will  also use new deep learning algorithms to help improve model estimation  and positioning results.


Team Name

GoD


Team members

Kai Luo, Ziyao Ma, Yanbiao Gao, Jizhou Wang, Deming Zhu, Yuqi Huo, Tianbao Pan, Xudong Song


Affiliation

Beijing University of Posts and Telecommunications


Description

According to the provided information of track 8, we proposed a  scheme to obtain the high-accuracy position based on the combination of  UL-TDOA and AI localization methods.

Our scheme can be divided  into four major parts. The first three parts show the preparatory works  including preprocessing of training data, machine learning based  location estimation and outlier calibration. The last part shows the  process of the final test, which includes: 1. Use the trained model to  get a tentative result; 2. Outlier detection and calibration; 3. Give  the final result.


Team Name

ISCAS


Team members

Chang Su, Fusang Zhang, Beihong jin


Affiliation

Institute of Software, Chinese Academy of Sciences


Description
5G positioning will be part of our  lives and it will enable diversified applications in all walks of life. A  large number of application scenarios such as the Internet of Vehicles,  autonomous driving, smart manufacturing, smart logistics, drones, and  asset tracking have higher requirements for positioning capabilities.  Excessive errors may lead to poor user experience or other problems.  Therefore, we should enhance network positioning technology to improve  5G positioning accuracy.
The challenges we face in Track 8 are the  features of raw data provided are inaccurate. These features are the  results of processing and it’s so hard to infer more information from  these features. What’s more, there exists a mixture of LOS paths, weak  LOS paths, and NLOS paths. And there are existing timing errors among  the receivers in TRPs, called time alignment errors (TAEs). The TAEs of  the TRPs are unknown and different in different datasets.