Analysis of key performance indicators of a 4G LTE network based on experimental data obtained from a densely populated smart city

Key performance indicator (KPI) data provide candidate information required for effective network planning, performance analysis and optimization. However, inadequate KPI data could limit efficient network planning leading to escalating operational cost, and this could adversely affect the subscribers of the network. To this end, this article presents radio frequency (RF) measurements and evaluation of KPIs taken at 1876.6MHz with a bandwidth of 10MHz, for an operational 4G LTE network in Nigeria. The measurements campaign specifically examine the behaviour of the RSRP, RSRQ, RSSI, SINR, PCC PHY DL Throughput, and the PDCP DL Throughput. Huawei Technologies Modem E392 was used for the propagation measurements, and RF measurements cover three evolved node base stations (eNodeBs) with average heights of 25 m. The geographical coordinates of the sites are as follows: Site 1 (Latitude 6.43543333; Longitude 3.44539667), Site 2 (Latitude 6.55639500; Longitude 3.36693333), and Site 3 (Latitude 6.51879500; Longitude 3.39911000). The E392 4G (LTE) Modem is capable of propagation measurements at the various LTE frequency bands, enables LTE download Speed of 100 Mbit/s, supports LTE upload Speed of 50 Mbit/s, utilizes LTE 2x2 MIMO (Multiple Input Multiple Output), and supports 64QAM (Quadrature Amplitude Modulation). The Drive Test (DT) Software version-Genex prove V16, and Genex Assistance V16 were deployed, and the test car carried a test terminal station, a GPS, a Windows supported Computer, and the accompanying drive test system. The test vehicle was driven such that it considered the actual road traffic conditions at a relatively medium speed of up to 30km/h with uniformity thereby reducing possible Doppler effects. Terminal connection was established, and data download services was started (using file transfer protocol - ftp, a drive test software, which has the function to download a large file of around 20GB). Thereafter, the download simultaneous file downloading limit was set to 5 files (such that 5 files can be downloaded simultaneously with quality download speed). When connection drops, simultaneous connection was re-established using the ftp software, and drive test was carried out within a planned cluster on a bright and sunny day. Statistical descriptions and probability distribution functions of the KPI data is reported and interdependence amongst the KPIs are presented to ease understanding of the interrelationships among the tested KPIs. The data reported would find useful applications in RF planning, radio channel measurements and modelling, feasibility studies and formulation of appropriate regulatory policies for wireless communication systems. Network operators could leverage on the data for appropriate KPI analyses, radio resources management, and research and development.

and Genex Assistance V16 were deployed, and the test car carried a test terminal station, a GPS, a Windows supported Computer, and the accompanying drive test system. The test vehicle was driven such that it considered the actual road traffic conditions at a relatively medium speed of up to 30km/h with uniformity thereby reducing possible Doppler effects. Terminal connection was established, and data download services was started (using file transfer protocol -ftp, a drive test software, which has the function to download a large file of around 20GB). Thereafter, the download simultaneous file downloading limit was set to 5 files (such that 5 files can be downloaded simultaneously with quality download speed). When connection drops, simultaneous connection was re-established using the ftp software, and drive test was carried out within a planned cluster on a bright and sunny day. Statistical descriptions and probability distribution functions of the KPI data is reported and interdependence amongst the KPIs are presented to ease understanding of the interrelationships among the tested KPIs. The data reported would find useful applications in RF planning, radio channel measurements and modelling, feasibility studies and formulation of appropriate regulatory policies for wireless communication systems. Network operators could leverage on the data for appropriate KPI analyses, radio resources management, and research and development.  The key performance indicator (KPI) data reported in this article were collected in and around three eNodeB sites with the following coordinates; Site 1 (Latitude 6.43543333; Longitude 3.44539667), Site 2 (Latitude 6.55639500; Longitude 3.36693333), and Site 3 (Latitude 6.51879500; Longitude 3.39911000), located in one of Africa's fastest growing smart city, Lagos, Nigeria. Data accessibility A detailed datasets on the measured KPIs taken at 1876.6MHz with a 10MHz bandwidth, of a functional 4G LTE network is provided as a supplementary file attached to this article in a spreadsheet format for easy accessibility and data reusability.

Data
Wireless communication data provide useful information pertinent to the development of communication equipment, standards and specifications, conducting high-level feasibility studies during initial deployment of telecommunication infrastructure, and providing accurate evaluation of the quality of service (QoS) [12] in order to enhance the quality of user experience (QoE) [13]. Generally, wireless communication systems are designed to transfer data from a source to a destination (from the transmitter to the receiver). As wireless systems continue to grow and evolve to accommodate upward scaling traffic requirements following the rapid deployment of 4G LTE networks and the evolving 5G and beyond wireless systems [2,5,14], analysis of the key performance indicators increasingly becomes a concern. Toward this end, the need to critically examine and evaluate the KPIs of an operational 4G LTE network becomes imperative. This is considered highly important due to the enormous benefits such data provide; useful information about the performance of the network in real time, and present a suitable platform to furnish improvement initiatives on the existing network structure in terms of coverage and capacity [15,16]. Finally, the data could aid the development of advanced modulation techniques [17e19], and foster the development of energy efficient wireless communications systems [20e22].
In this article, analysis of some selected KPIs of an operational 4G LTE network is presented. The tested KPIs include the RSRP, RSRQ, RSSI, SINR, PCC PHY DL Throughput, and the PDCP DL Throughput. These KPIs were measured at a 4G LTE frequency of 1876.6MHz with 10MHz bandwidth. The extensive RF measurements span a propagation distance of up to 2km, and measured KPIs were extracted and analysed in IBM SPSS Statistics and MATLAB.
The KPIs derived from the experimental data are briefly described as follows. First, the aerial view and the geographical coordinates of the measurements environment are as shown in Figs. 1e2, respectively. The trajectories of 4G LTE RSRP, RSRQ, SINR, and PCC PHY Throughput performance distributions are as shown in Figs. 3e6, respectively. The specific KPI information are presented in Figs. 7e12. Specifically, the RSRP measured at Sites 1e3 is given in Fig. 7, and Fig. 8 shows the RSRQ measured at Sites 1e3. Fig. 9 represents the RSSI measured at Sites 1e3, Fig. 10 gives the SINR measured at Sites 1e3, and Fig. 11 presents the PCC PHY DL Throughput measured at Sites 1e3. Finally, Fig. 12 shows the PDCP DL Throughput measured at Sites 1e3.
The statistics of the measured KPIs are given in Tables 1e9. More specifically, Table 1 presents statistics of measured RSRP, RSRQ, and the RSSI at Site 1. Table 2 gives the statistics of measured SINR, PCC PHY DL Throughput, and the PDCP DL Throughput at Site 1. Table 3 represents the statistics of measured RSRP, RSRRQ, and the RSSI at Site 2. Table 4 depicts the statistics of measured SINR, PCC PHY

Value of the Data
The experimental data reported in this article will enhance further research in the field of wireless communications engineering, especially in the area of radio channel measurements and key performance indicator analyses in dense urban propagation environments [2e4]. The data will also be of immense benefits to: 1) Radio Network Engineers for assessing and determining the optimal location of base stations (BSs), radio channels and radio coverage estimations, and capacity improvements. 2) Radio Frequency Planning Engineers for radio frequency planning, frequency assignments and network optimization, drive testing and optimal allocation of radio resources, and quality of service (QoS) analyses. 3) Regulatory and Compliance Engineers can also leverage on the data to provide suitable KPI benchmarks for mobile network operators [5e7].
The KPI data will provide further insights and development of experiments in the area of radio network design, development and validation of high precision propagation models for accurate prediction of pathloss in environments where radio signals are severely impacted by multi-scattering attenuation under different environmental conditions [7,8].
The data could also find additional use as candidate materials for class room studies (testing and validating theoretical and simulation results) [9e11].
DL Throughput, and the PDCP DL Throughput at Site 2. In addition, Table 5 presents the statistics of measured RSRP, RSRQ, and the RSSI at Site 3, whereas, Table 6 gives the statistical analysis of the measured SINR, PCC PHY DL Throughput, and the PDCP DL Throughput at Site 3. Furthermore, Table 7 gives a comparison of the measured RSRP and the RSRQ at Sites 1e3, and a comparison of the measured  RSSI and the SINR at Sites 1e3 is given in Table 8. Last, Table 9 shows a comparison of the measured PCC PHY DL Throughput, and the PDCP DL Throughput at Sites 1e3. The probability distribution of the KPIs observed are given in Figs. 13e18. Notably, Fig. 13 illustrates the probability density of the measured RSRP at Sites 1e3. Fig. 14 gives the probability density of the measured RSRQ at Sites 1e3, and Fig. 15 provides the probability density of the measured RSSI at Sites 1e3. In the same vein, Fig. 16 reports the probability density of the measured SINR at Sites 1e3, whereas, Fig. 17 represents the probability density of the measured PCC PHY DL Throughput at Sites 1e3. Finally, Fig. 18 presents the probability density of the measured PDCP DL Throughput at Sites 1e3.

Experimental design, materials, and methods
The equipment used for measurements is the Huawei Modem E392. The E392 4G (LTE) Modem offers flexibility in RF measurements and post processing of measurements data. The equipment can be used for propagation measurements at various LTE frequency bands, and supports a LTE download Speed of 100 Mbit/s, while the LTE upload Speed supported is up to 50 Mbit/s. Furthermore, the device supports LTE 2x2 MIMO and 64QAM (Quadrature Amplitude Modulation). The Drive Test (DT) Software version-Genex prove V16, and Genex Assistance V16 were selected and carefully connected and assembled in the DT car for seamless propagation measurements. The drive test car carried the test terminal station, the GPS equipment, and a personal computer (PC), and the associated drive test system.              In order to achieve quality results, the test vehicle was driven such that it considered the actual road traffic conditions at medium speed of up 30km/h with uniformity. This helps to reduce the possible impacts of Doppler effects. Afterwards, the terminal connection was established, and data download services started using file transfer protocol -ftp, a drive test software, which has the function to download a large file around or up to 20 giga bytes (GB). Thereafter, the download simultaneous file downloading limit was set to 5 files (such that 5 files can be downloaded simultaneously without significant computational cost especially on the baseband processing unit). When connection drops, simultaneous connection was re-established using the ftp software, and drive test was carried out within a planned cluster located in the geographical coordinates of the measurements environment. For data post processing, MATLAB 2018a, a product of Mathworks Incorporated, and the IBM Statistical tool (SPSS) version 24 were used.