An Accuracy and Real-Time Commercial Localization System in LTE Networks

Abstract

In this paper, an ARTLoc (accuracy and real-time commercial localization) system is proposed, which is applicable for positioning the Minimization of Drive-Tests(MDT) data in the long-termevolution(LTE) cellular communication network system. This system utilizes a collaborative filtering algorithm to ensure the robustness and accuracy of the fingerprint database when both base station position and GPS MDT data are abnormal in commercial communication service provider(CSP) MDT localization scene. Then, a fast matching KNN algorithm including coarse matching and fine matching is proposed to improve the location efficiency and reduce the location cost under the premise of ensuring high location accuracy. The results of experiments conducted in an in-service LTE network using more than 1000 LTE base station demonstrate that the proposed technique yields a location accuracy of 65.7 meters(@67%) with DT data and 62.1 meters(@67%) with massive GPS MDT data(over 10 million), which provides at least 184.2% and 145.2% enhancement in accuracy respectively compared to the traditional technology. In addition, our system provides a more than 7 times improvement in location efficiency compared to traditional technology. This proposed localization technique is applicable in network optimization and Operation and Maintenance(O&M) to assist CSP to reduce their operating expense(OPEX) by positioning those MDT data without GPS location.

An Accurate and Real-Time Commercial Indoor Localization System in LTE Networks

“…Some other fingerprint based localization systems take into account the commercial costs, such as CellSense [19], NBL [24], ARTLoc [8] and DeepLoc [25]. These systems utilize GNSS MDT data to generate the FD in offline phase, and then complete the locating of non-GNSS data in online phase.…”

“…Since traditional DT data is replaced by GNSS MDT data to build FD, localization techniques used in these systems possess the characteristics of low commercial cost, which makes these systems applicable on a large scale. However, due to the severe deterioration of GNSS location performance in indoor environment, the FD built by GNSS MDT data is mainly distributed outdoors [8], [19], which will result in indoor MDT data being located outdoors. In this case, the indoor location accuracy obtained by these systems will be seriously reduced, which makes many indoor applications unable to be used.…”

“…Traditional localization scene can be categorized into two groups based on request initiator and data sets used in the methods: UE and CSP MDT localization scene respectively [8]. For UE localization scene, localization methods running in Google Maps, Amap or other navigation applications (Apps), are mainly initiated by UE.…”

Enhanced MDT-Based Performance Estimation for AI Driven Optimization in Future Cellular Networks

“…Moreover, the impact of quantization or sparse measurements was not the focus of this work. Authors in [21] propose a system that utilizes a collaborative filtering algorithm for improving robustness and accuracy of the MDT reports when both base station position and GPS MDT data are abnormal. After first estimating the position of base station, the abnormal GPS MDT data is detected and eliminated with the help of estimated base station position.…”

“…||.|| F denotes the Frobenius norm. ⌘ is a regularization parameter in the objective function and is shown in [61] that the solution of the problem in (21) converges to that of (19) as ⌘ ! 1.…”