A feasibility study of 5G positioning with current cellular network deployment

Abstract

This research examines the feasibility of using synchronization signals broadcasted by currently deployed fifth generation (5G) cellular networks to determine the position of a static receiver. The main focus lies on the analysis of synchronization among the base stations of a real 5G network in Milan, Italy, as this has a major impact on the accuracy of localization based on time of arrival measurements. Understanding such properties, indeed, is fundamental to characterize the clock drifts and implement compensation strategies as well as to identify the direct communication beam. The paper shows how the clock errors, i.e., inaccurate synchronization, among 5G base stations exhibit a significant bias, which is detrimental for precise cellular positioning. By compensating the synchronization errors of devices' clocks, we demonstrate that it is in principle possible to localize a static user with an accuracy of approximately 8-10 m in non-obstructed visibility conditions, for urban and rural scenarios, using the deployed 5G network operating at 3.68 GHz and relying on broadcast signals as defined by 5G Release 15 standard. This work has been funded by the European Space Agency (ESA) Navigation Innovation and Support Program (NAVISP) Element 2 pillar which aims at improving the competitiveness of the industry of the participating States in the global Positioning, Navigation and Timing (PNT) market.

Figure 1

SSB patterns as described by 3GPP Release 15 standard.

Figure 2

Representation of the urban scenario selected for the experiments and composed of three 5G sites (red squares), each with 3 TRPs (red icons) with indicated PCI identifiers, and 5 UE measurement points (blue icons). The power map is computed with MATLAB ray-tracing software at ground level.

Figure 3

Three sites of the urban scenario in Fig. 2, with related TRPs. For each TRP, the color of PCI number indicates the visibility condition: green color is for LOS, red color is for NLOS.

Figure 4

Distribution of the synchronization error in meters, i.e., range bias, for the TRPs in visibility (identified by their PCIs) for each of the 8 SSB beams.

Figure 5

Distribution of the synchronization error in meters, i.e., range bias, for the TRP in non-visibility (identified by their PCIs) for each of the 8 SSB beams.

Figure 6

Normalized autocorrelation function of the range biases of TRP with PCI 252 for each SSBs.

Figure 7

Evolution of range bias (in m) over time for PCI 252, 411 and 383 with corresponding 2D representation of received power per beam. Colors indicate the value of measured received power in dBm.

Figure 8

Simulated representation of the rural scenario composed of 12 TRPs (red icons) identified by the indicated PCI number and a static measurement point (blue icon). The power map is computed with MATLAB ray-tracing software at ground level.

Figure 9

Urban scenario: (a) scatterplot of the position estimates in case of range correction (green squares), without range correction (blue squares) and EKF. The ellipses are obtained with 95% of confidence level. Black squares and solid black lines represent the mean position estimates and the error bias, respectively, while the CRB is highlighted with orange line. (b) CDF of the UE positioning error.

Figure 10

Rural scenario: (a) scatterplot of the positions estimates in case of range correction (green squares), without range correction (blue squares) and with EKF. The ellipses are obtained with 95% of confidence level. Black squares and solid black lines represent the mean position estimates and the error bias, respectively, while the CRB is highlighted with orange line. (b) CDF of the UE positioning error.

Figure 11

SSB beam sweeping along different directions. The receiver, i.e., R&S TSMA6 scanner, gathers both direct and reflected signals, experimenting a multipath delay. $\mathrm{\Delta }{t}_i$ and $\mathrm{\Delta }{t}_{\text{UE}}$ are the clock offset of the TRP i and UE, respectively. On the contrary, the distance between the TRP i and UE are indicated with $d_{i-\text{UE}}$.

Figure 12

The radial axes represents the range bias in meters while the color indicates the received power for each measure. (a) and (b) show the beams in 3D and 2D, respectively.