Indonesian J our nal of Electrical Engineering and Computer Science V ol. 42, No. 1, April 2026, pp. 105 ∼ 114 ISSN: 2502-4752, DOI: 10.11591/ijeecs.v42.i1.pp105-114 ❒ 105 Corr elation-based assessment of 4G L TE netw ork perf ormance during rainfall e v ents in tr opical r egions Ngozi C. Eli-Chukwu, Uma Uzubi Uma, Handel Emezue, Ogechi Akudo Nw ogu, Ogah E. Oga, Calister N. Ogbonna-Mba, Samuel I. Ezichi Department of Electrical/Electrnoics Engineering, Ale x Ekwueme Federal Uni v ersity Ndufu Alik e, Ebon yi State, Nigeria Article Inf o Article history: Recei v ed No v 12, 2025 Re vised Jan 19, 2026 Accepted Mar 4, 2026 K eyw ords: F ourth-generation Recei v ed signal strength indicator Reference signal recei v ed po wer Reference signal recei v ed quality Signal-to-interference noise ratio ABSTRA CT This paper presents a performance e v aluation of a fourth-generation (4G) cel- lular netw ork under adv erse weather conditions in a tropical re gion. While the impact of rainf all on frequencies abo v e 10 GHz is well documented, this study addresses the research g ap concerning 4G L TE performance (sub-6 GHz ) in high-precipitation en vironments such as Nigeria. Using a dri v e-test approach with TEMS In v estig ation softw are (v16.3), measurements were collected o v er 48 days between July and September 2025 along a x ed 15 km route in the Lagos metropolis on the MTN Nige ria netw ork. Samples were recorded at 1- second interv als. F our critical k e y performance indicators (KPIs)—reference signal re cei v ed po wer (RSRP), reference signal recei v ed quality (RSRQ), signal- to-interference-plus-noise ratio (SINR), and recei v ed signal strength indicator (RSSI)—were analyzed to determine their inuence on the netw ork performance inde x (NPI). Correlation analysis re v ealed tha t while RSRP e xhibits no sig- nicant correlation with NPI during rainf all ( r s = 0 . 009 ), SINR and RSRQ demonstrate strong positi v e correlations ( r s = 0 . 828 and r s = 0 . 824 , respec- ti v ely). Despite these high correlations, a v erage performance v alues remained lo w (mean SINR = 23.72%), indicat ing signicant rain-induced de gradation. These ndings pro vi de a no v el empirical basis for the de v elopment of weather - a w are adapti v e algorithms in tropical 4G netw ork deplo yments. This is an open access article under the CC BY -SA license . Corresponding A uthor: Uma Uzubi Uma Department of Electrical/Electrnoics Engineering, Ale x Ekwueme Federal Uni v ersity Ndufu Alik e Ebon yi State, Nigeria Email: uma.uzubi@funai.edu.ng 1. INTR ODUCTION The global dependence on wireless communication has intensied the demand for high quality of service (QoS). Ho we v er , wireless channels are highly susceptible to en vironmental f actors, particularly hea vy rainf all in tropical re gions. In recent years, there has been a gro wing global dependence on wireless commu- nication systems [1]. The demand f or cellular netw ork services continues to increase at a geometric rate [2], [3], leading to an e xponential e xpansion in wireless netw ork usage. The emer gence of the internet of things (IoT), sensor technologies, and other connected media has further intensied the demand for reliable wireless infrastructure. These systems e xhibit v arying sensiti vity le v els and therefore require optimal QoS for ef cient operation. Presently , wireless netw orks form an indi spensable component of socio-economic acti vities across multiple s ectors, becoming inte gral to daily human interaction and b usiness processes [4]. Consequently , ensur - ing the optimal performance of wireless systems is essential for maintaining operational continuity and service J ournal homepage: http://ijeecs.iaescor e .com Evaluation Warning : The document was created with Spire.PDF for Python.
106 ❒ ISSN: 2502-4752 sustainability . T o achie v e this, netw ork operators must identify and mitig ate k e y performance indicators (KPIs) that adv ersely af fect performance. Man y of these impairments are naturally induced, as wi reless channels are highly susceptible to en vironmental f actors such as rainf all and other adv erse weather conditions [5]. Such conditions de grade signal quality and netw ork reliability . Therefore, e v aluating netw ork performance under these circumstances is crucial for identifying KPIs that signicantly impact service quality . This study focuses on assessing the 4G netw ork KPIs that inuence performance during periods of hea vy rainf all. Rainf all is a seasonal natural occurrence globally and is usually hea vy in countries located in the tropical re gions of the w orld. Ov er the years, climate change due to global w arming has led to increased ooding, hurricanes, and other adv erse weather e v ents, gi ving rise to hug e economic losses. The losses also af fect telecommunication f acilities, especially in de v eloping countries. Therefore, there is a need to e v aluate the performance indicators that af fect cellular netw orks during rainf all to impro v e netw ork quality . Since cellular netw orks are global, an y ef fect on cellular netw ork performance is critical. During the rain y season, the radio en vironment of cellular netw orks e xperiences a decline that mak es accessing and retaining services rendered by the netw ork operator cumbersome. Since w ater conducts electric- ity , radio w a v es can be block ed and reected during the rain y season. The ef fect on communication netw orks is usually more predominant in count ries in tropical clime such as Nigeria, where the rate of rainf all could be as high as 200 cm annually . This will certainly af fect the performance of cellular netw orks. This ef fect is not limited to tropical climes, as studies ha v e sho wn a corresponding decline in cellular netw ork performance in temperate re gions. Rain attenuation results in outages that af fect the quality of the signal , performance of cellular netw orks, link a v ailability , de viation of the recei v ed signal, and restriction of the path length of radio communication signal, thereby limiting the utilization of higher frequencies for the line-of-sight micro w a v e links [6]. Eli-Chukwu and Onoh [7] present the ef fect of rainf all on link quality in outdoor forest deplo yment and found that rainf al l decreases link performance by 62 percent to the link with pack et reception and 30 per - cent to the link ha ving belo w 50 percent reception. Thi s is caused by rainf all and w ater sitting on the nodes e v en after the rain. Al-samman et al. [8] considered some KPIs that are af fected during rainf all and concluded that rainf all af fects signal strength and equally observ ed a signicant drop in the recei v ed signal strength indi- cator (RSSI). This corroborates the ndings of Y aghoubi et al. [9] that recei v ed signal strength is susceptible to radio propag ation conditions, during periods of high precipitation. F ong et al. [10] authors presented the results of a study carried out in southeast Nigeria and concluded that adv erse weather conditions (rain y and harmattan season) af fect the performance of cellular communication. The drop call rate w as as high as 12.89% as ag ainst the standard practice of 2% and the block call rate w as 8.76% a s ag ainst the re gulatory of 2%. Ov er time, re search has been done to emphasize the import ance of the 4G netw ork and the ef fect of v arious netw ork conditions on the cellular netw ork. F ourth-generation broadband cellular netw orks ha v e re v olutionized the mobile industry . It can hold video conferencing, high-denition mobile tele vision, better circuit calls and man y others. T o do this requires lo w latenc y as most of the products are real-time interaction based. F or instance, video conferencing and online g aming are real-time interactions, so an ything that interrupts communication af fects its performance. Rainf all is seen as one of the natural occurrences that af fect the quality of the cellular netw ork. W ith the high demand for data and real-time interaction which are some of the k e y usage cases of 4G netw orks, mitig ating f actors that will af fect its performance are v ery crucial. V arious research conducted based on the impact of rainf all on electromagnetic w a v es considered only the electromagnetic w a v es with frequencies greater than 10 GHz since the impact of rainf all on the electromag- netic w a v e attenuation is reduced at lo wer frequencies [11]. F an et al. [12] presented the ef fect of rain atten- uation on S-band terrestri al links and found that hea vy rainf all attenuates the communication s ignals thereby af fecting the netw ork performance ne g ati v ely . Se v eral f actors contrib ute to signicant link performance de gra- dation. Most se v ere is the inuence of the adv erse atmospheric conditions that frequently appear , thus making the design of strongly connected netw orks a demanding issue. A good netw ork link should successfully combat fog, rain, and sno w , and deri v e analytical e xpressions for the node isolation probability , assuming a suitable path loss model [13] and equally e xamined the accurac y of K u-band Ocean wind scatter meters which are af fected by v arying de grees of rain. The authors state that horizontal polarization measurements are more sensiti v e to rain than v ertical polarization. Sensiti vity to rain v aries dramatically with wind speed, and the additional backscatter due to rain o v ershado ws the rain-related attenuation. Thiag arajah et al. [14] state that netw ork performance de gradation is caused by long-lasting channel uctuations such as rain f ading, therefore required treatment carefully . A rain detection algori thm utilizing both temporal and spatial correlation of link Indonesian J Elec Eng & Comp Sci, V ol. 42, No. 1, April 2026: 105–114 Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752 ❒ 107 status, aiming at ef ciently distinguishing between long-term and short-term channel f ading can be used to mitig ate the ef fects of rain. Rain f ading may cause some links to f ail, thereby threatening the connecti vity of the netw ork such as in millimetre w a v elength mesh netw orks. Cellular netw ork co v erage is greatly af fected by rainf all. Rain attenuation on the radio channel in a hea vy rainf all re gion for wideband signals is in the 10–40 GHz frequenc y range. A lo wer frequenc y of 10 GHz is more af fected by multipath due to a longer w a v elength; it is still optimal in a hea vy rainf all re gion due to high losses associated with higher frequenc y signals. In addition, for a path of o v er 1.5 km, the dif ference in horizontal and v ertical polarization is signicant [15]. One of the natural phenomena that signicantly inuences the performance of communicati on systems is rain attenuation. This is particularly true when the inuence of rain is put into the conte xt of point-to-multipoint, x ed cellular radio netw orks, in which spatial statistics of rain attenuation are as important as single-point temporal statistics [16]. P aul et al. [17] in v estig ated the impact of rain attenuation on wireless communication sys tems and demonstrated that rainf all causes signicant signal de gradation due to absorption and scattering, leading to increased path loss and reduced link reliability . Their ndings conrmed that higher rain intensity results in noticeable deterioration of signal quality , particularly af fecting error performance in cellular netw orks. Xu et al. [18] in v estig ated the impact of rainf all on cellular communication links using smartphone-collected measure- ments, re v ealing signicant v ariations in signal metrics under rain y conditions. Michel et al. [19] in v estig ated the impact of rain on L TE signal propag ation in connected car scenarios using lar ge-scale 3GPP minimization of dri v e test (MDT) measurements collected along the Italian A22 Motorw ay . By comparing L TE recei v ed signal recei v ed po wer (RSRP) under rain y and dry conditions, the study sho wed that wet en vironments signif- icantly alter electromagnetic propag ation e v en at L TE frequenc y bands belo w 3 GHz. The authors highlighted that rain ef fects on cellular propag ation are critical for connected and autonomous mobility , particularly with the deplo yment of higher -frequenc y 5G bands that are more susceptible to rain attenuation. Sheu et al. [20] studied the impact of tropospheric conditions on cellular signal propag ation, sho wing that temperature, humidity , air pressure, and precipitation cause attenuation, scatteri ng, and ducting, which de grade netw ork per formance. Using KPIs such as CSSR, HOSR, and RSCP , the y found that humidity and rain signicantly reduce signal quality , while ducting can enhance co v erage. The study recommends real-time atmospheric monitoring, adapti v e po wer control, and frequenc y tuning to impro v e netw ork reliability and QoS. P atra and Mitra [21] in v estig ated rain f ading in tropical re gions, where hea vy precipitation reduces RF s ignal po wer and de grades radio system performance. T o mitig ate rain-induced signal loss, the authors implemented a model using frequenc y di v ersication, adapti v e po wer control, and w a v eform techniques. Their e xperiments, conducted at 26 GHz and 38 GHz, demonstrated that frequenc y di v ersication ef fecti v ely im- pro v es system performance. Comparisons with e xisting rain attenuation models v alidated the proposed model across dif ferent distances, frequencies, and ele v ation angles. Abdullahi and Abdulhamid [22] e v aluated the performance of 4G/L TE netw orks in Nort h-Central Nigeria using dri v e test measurements across major mobile netw ork operators. KPIs such as netw ork speed, latenc y , pack et loss, co v erage, and signal po wer were analyzed using propag ation models including Oku- mura–Hata and W alsch–Ik e g ami. The results sho wed that MTN achie v ed the best o v erall performance, fol- lo wed by Airtel, while GLO and 9Mobile e xhibited comparati v ely lo wer metrics. The study emphasizes the v alue of comprehensi v e KPI-based e v aluation for impro ving netw ork quality and user e xperience. Despite the challenges f acing cellular netw ork pro viders during adv erse weather condit ions, the re g- ulatory body and customers still w ant the set quality threshold to be met by the service pro viders. Hence, there is a constant need for netw ork optimization, particularly during the rain y season as cellular signals are at their weak est during that period. Most studies were carri ed out on its impact on 2G and 3G netw orks. Ho we v er , it is important to note that dif ferent netw ork parameters ha v e dif ferent parameters that are af fected during adv ers e weather conditions. Little research has been conducted on the adv erse weather ef fects on 4G netw orks. This paper studies the performance e v aluation of parameters impacting cellular netw ork QoS during the rain y season, the v arious parameters that inuence performance and ho w the y af fect the o v erall netw ork performance. The authors attempt to in v estig ate the netw ork parameters that are most af fected due to rainf all in 4G netw orks. F our netw ork performance indicators were monitored; RSRP , referenced signal recei v ed quality (RSRQ), signal-to- interference noise ratio (SINR), and recei v ed signal strength indicator (RSSI). This paper presents results from outdoor monitoring of the performance of these netw ork parameters for forty-eight weeks within three months during hea vy rainf all in Nigeria. The quality of the RSRP , RSRQ, SINR, and RSSI w as check ed whene v er there w as hea vy rainf all. Correlations were performed with the benchmark ed netw ork data Corr elation-based assessment of 4G L TE network performance during r ainfall ... (Ngozi C. Eli-Chukwu) Evaluation Warning : The document was created with Spire.PDF for Python.
108 ❒ ISSN: 2502-4752 to ascertain each parameter’ s performance ag ainst netw ork performance indicator benchmark ed data. Problem Statement and Research Gap: While pre vious research has focused e xtensi v ely on rain at- tenuation in 2G and 3G net w orks or frequencies e xceeding 10 GHz, little research has been conducted on the specic vulnerabilities of 4G L TE parameters during hea vy precipitation. Furthermore, the relationship be- tween signal po wer and actual link quality during storm e v ents is often o v erlook ed. This study aims to ll this g ap by e xplicitly e v aluating the NPI—a composite industry-standard metric representing the o v erall reliability and accessibility of the radio link—ag ainst indi vidual KPIs. The main objecti v e of this paper is as follo ws: − T o identify which 4G KPIs are most af fected by hea vy rainf all. − T o quantify the correlation between these KPIs and the NPI. − T o propose an optimization, focus for netw ork operators to sustain reliability during adv erse weather . 2. METHOD The dri v e-test methodology utilized TEMS In v estig ation softw are (v16.3) insta lled on a high- performance laptop, a 4G-enabled TEMS mobile de vice, and a high-precision GPS recei v er . Field measure- ments were conducted across t he Lagos metropolis on the MTN Nigeria netw ork (the sole operator for this study) o v er 48days from July to September 2025. Rainf all Characterization: Measurements were cate gorized during ”Hea vy Rainf all” e v ents, typical of the Nigerian tropical rain y season. While specic rain-g auge sensors were deplo yed on the v ehicle, data collection w as synchronized with re gional meteorological reports indicating precipitation intensities e xceeding 50 mm/h. The dri v e-test route co v ered a mixture of urban can yons and open residential areas to ensure a di v erse sampling of multipath f ading and rain scattering. 3. RESUL TS AND DISCUSSION The analysis focused on four k e y parameters and their relationship with the NPI. A nota b l e nding is the relationship between lo w performance v alues b ut high c o r relations. F or instance, the SINR mean w as observ ed at a lo w 23.72%, yet it maintained a high correlation ( r s = 0 . 828 ) with NPI. This indicates that while rainf all consistently suppresses signal quality to a lo w baseline, the o v erall netw ork performance (NPI) remains e xtremely sensiti v e to the minor uctuations of SINR during these periods. T able 1, Correlation Coef cients ( r s ) for KPIs during Rainf all with the standard − SINR: 0.828 (Strong) − RSRQ: 0.824 (Strong) − RSRP: 0.009 (Insignicant) 3.1. Netw ork perf ormance index corr elation model T o establish the relationship betwee n the NPI and selected KPIs—namely RSRP , RSRQ, SINR, and RSSI—data obtained from a 48-day dri v e test were analyzed. Based on the computed dataset, a correlation model w as de v eloped to determine the specic KPIs inuencing 4G netw ork performance during the rain y season. This relationship w as deri v ed by performing an independent correlation analysis between the NPI and each KPI. The de gree of correlation indicates whether a gi v en KPI e x erts a positi v e or ne g ati v e inuence on o v erall netw ork performance. The correlation model is mathematically e xpressed using Spearman’ s rank correlation coef cient as: r s = 1 − 6 P d 2 q n ( n 2 − 1) (1) where n represents the number of observ ations for each netw ork v ariable, d q denotes the rank dif ference be- tween the tw o v ariables under comparison (e.g., RSRP and NPI), and r s is the correlation coef cient, which ranges from − 1 to +1 , indicating ne g ati v e and positi v e relationships, respecti v ely . T o further e v aluate the mean performance of each KPI relati v e to the netw ork performance benchmark, the e xpression in (2) w as applied. This f o r mulation enables proper assessment of the percentage performance of the co v erage area with respect to the NPI: µ q = 1 n n X i =1 q i , (2) Indonesian J Elec Eng & Comp Sci, V ol. 42, No. 1, April 2026: 105–114 Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752 ❒ 109 where µ q represents the mean percentage v alue of each KPI (RSRP , RSRQ, SINR, and RSSI), and q i denotes the indi vidual percentage measurements for each observ ation. T able 1. Daily percentage distrib ution of KPI thresholds and NPI during rainf all Month Day RSRP ≥ -95 dBm (%) RSRQ ≥ -14 dB (%) SINR ≥ 10 dB (%) RSSI ≥ -50 dBm (%) NPI July 1 68.75 60.33 49.67 48.18 2.84 2 62.33 58.00 65.67 40.91 2.84 3 64.25 58.67 51.33 43.00 2.72 4 68.08 58.00 35.33 46.82 2.60 5 68.58 57.00 31.33 47.18 2.55 6 67.75 56.00 40.67 46.45 2.64 7 66.83 57.00 47.00 45.82 2.71 8 68.67 52.33 16.67 46.18 2.30 9 63.67 50.33 1.33 40.45 1.95 10 66.75 46.67 1.00 43.27 1.97 11 66.00 51.67 7.00 43.18 2.10 12 67.58 47.00 8.67 41.91 2.06 13 62.25 52.33 15.67 39.36 2.12 14 62.92 56.67 28.67 41.18 2.37 15 62.58 57.00 47.67 41.09 2.54 16 64.92 56.33 38.67 43.45 2.54 17 74.75 46.67 6.33 50.64 2.23 18 73.67 46.00 7.67 49.82 2.21 19 73.83 46.00 3.67 50.09 2.17 20 73.75 48.00 4.00 50.55 2.20 21 74.42 48.33 6.33 51.45 2.26 22 40.08 90.33 16.33 52.18 2.49 23 74.92 56.67 26.00 54.00 2.64 24 75.83 56.00 22.00 54.91 2.61 August 25 74.42 56.67 32.67 53.55 2.72 26 75.42 56.67 34.00 54.64 2.76 27 73.83 58.33 46.00 53.45 2.90 28 77.67 57.33 33.00 57.45 2.82 29 79.33 56.00 25.67 58.55 2.74 30 76.75 58.00 46.00 56.09 2.96 31 68.92 58.67 69.67 47.64 3.26 32 65.58 59.00 75.33 45.45 3.07 33 80.67 24.00 13.00 47.36 2.06 34 67.83 56.33 19.33 46.27 2.37 35 69.42 57.00 26.33 48.18 2.51 36 69.17 58.33 27.00 48.27 2.53 September 37 71.50 57.33 22.33 51.00 2.53 38 75.50 56.33 24.00 54.91 2.63 39 79.00 54.67 11.67 58.18 2.54 40 79.92 52.00 5.33 58.64 2.45 41 79.50 51.67 5.67 58.00 2.44 42 79.17 51.33 4.33 57.45 2.40 43 79.25 51.00 4.33 57.55 2.40 44 78.92 51.33 5.00 57.18 2.41 45 79.08 52.00 5.00 57.55 2.42 46 78.67 52.33 7.00 57.18 2.44 47 76.75 53.33 14.33 55.27 2.50 48 78.92 51.00 3.00 57.18 2.38 3.2. Analysis of netw ork KPIs RSRP is dened as the linear a v erage of the po wer contrib utions of the resource elements carrying reference signals in a 4G L TE netw ork. T o determine the relationship between RSRP and the NPI, the cor - relation coef cient w as computed using (1), yielding r s = 0 . 009 . This e xtremely lo w coef cient indicates a ne gligible relat ionship between RSRP and the NPI during the rain y season. Consequently , v ariations in RSRP ha v e little or no inuence on o v erall 4G netw ork performance under rainf all conditions, as illustrated in Figure 1. Furthermore, the mean RSRP performance during the rain y season, computed using (2), w as found to be 71.42%. This result suggests that rainf all has minimal impact on RSRP le v els, reinforcing the observ ation Corr elation-based assessment of 4G L TE network performance during r ainfall ... (Ngozi C. Eli-Chukwu) Evaluation Warning : The document was created with Spire.PDF for Python.
110 ❒ ISSN: 2502-4752 that signal strength alone is not a reliable indicator of netw ork performance de gradation during adv erse weather conditions. RSRQ analysis is the KPI that sho ws the quality of the recei v ed reference signal. It pro vides further hando v er criteria when the RSRP report is not suf cient. When the data from the co v erage w as subjected to a correlation test by applying (1), the correlation coef cient had a positi v e correlation of r s = 0 . 824 . This indicates a positi v e relationship with the netw ork performance. Further analysis as can be seen in Figure 2, sho ws that at v arious points where the RSRQ = − 14 dB ≥ 0% , the NPI rose. This corroborates the mean of 54.25% which indicates that the o v erall netw ork performance w as af fected. Figure 1. Comparing the RSRP and NPI Figure 2. Comparing the RSRQ and NPI SINR is a measure of the signal quality , interference, and noise quantity . It sho ws the strength of the required signal when compared to the system noise and interference. T o calculate the correlation coef cient with the NPI, the data from the logles were subjected to a correlation test using (1). The result indicates a correlation coef cient of SINR performance r s = 0 . 828 which sho ws a strong relationship with the o v erall netw ork performance when compare d to other netw ork KPIs. Additionally , comparing the NPI with the SINR, the mean (SINR 10dB) is 23.72 % which as depicted does not help the o v erall netw ork performance as sho wn in Figure 3. It w as also observ ed that in some weeks where the SINR w as 10dB 50%, the NPI > ≥ 2 . 8 . This is a clear indication that in adv erse weather conditions, the SINR is greatly af fected which suggests that signal quality within that period will be drastically af fected [23]–[25]. RSSI it measures the total recei v ed po wer including all interference and thermal noise. Applying (1) gi v es a correlation coef cient of r s = 0 . 176 . Applying (1) and (2) the RSSI gi v es the follo wing results: r s = 0 . 176 . The ndings re v eal an insignicant relationship bet ween the RSSI and the NPI. The performance of the RSSI in the rain y season does not af fect the o v erall netw ork performance. The mean (RSSI ≥ − 50 dBm ) is 50.19%. 3.3. Optimization model The approach of multiple linear re gressi on w as used to predict the NPI (y) using independent v ariables x 1 (RSRP), x 2 (RSRQ), x 3 (SINR), and x 4 (RSSI). The model is dened by (1) and (2): y = β 0 + β 1 x 1 + β 2 x 2 + β 3 x 3 + β 4 x 4 + ϵ (3) Based on the re gression coef cients ( β ), a unit increase in the foun da tional parameters leads to a 1.08- unit increase in NPI. Clarication of Priorities: Although SINR is the most strongly correlated with NPI, we recommend that opt imization ef forts be prioritized on RSRP and RSRQ. This is because SINR is a composite v alue highly af fected by e xternal noise and interference during rain, making it harder to control directly . By stabilizing RSRP and RSRQ through site-le v el tilt adjustments and po wer boosts, the more v olatile SINR naturally stabilizes, leading to the predicted 1.08-unit g ain in NPI. The approach of multiple linear re gression is used to predict the NPI (independent v ariable) using se v eral e xplainable (dependent) v ariables. It leads to a model that ts a linear equation to the observ ed data. Indonesian J Elec Eng & Comp Sci, V ol. 42, No. 1, April 2026: 105–114 Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752 ❒ 111 Let the dependent v ariable be y and the independent v ariables x j s, the model format sho wn in Figure 4. Y i = β 0 + β 1 x i 1 + · · · + β p x ip + e i (4) Y q = 0 . 003 + 0 . 012 x 1 + 0 . 012 x 2 + 0 . 013 x 3 + 0 . 013 x 4 (5) The signicant v alue for the independent v ariables is 0 and as such the y are all signicant in predict ing the NPI. But the RSRP and RSRQ most ne g ati v ely af fect the netw ork performance during the rain y season as the y both ha v e a slo wer impro v ement rate. Although optimization is necessary on all the parameters considered, more focus should be on the RSRP and RSRQ because a unit impro v ement in the RSRP and RSRQ will yield a 1.08-unit impro v ement on SINR and RSSI. Figure 3. Comparing the RSRQ and NPI Figure 4. Comparing the RSSI and NPI 4. CONCLUSION This study conrms that rainf all signicantly de grades 4G performance in tropical re gions. Signic ant T ak e-a w ay: F or rob ust netw ork performance, operators must mo v e be yond po wer -bas ed (RSRP) monitoring alone and implement weather -a w are algorithms that prioritize RSRQ and SINR stabil ization during storms. Globally , wireless netw orks, especially cellular systems ha v e become indispensable for communicati on. The increasing demand for seamless connecti vity across consumer and industrial domains underscores the need for reliable netw ork performance under all conditions, including se v ere rainf all. As climate change intensies en- vironmental disruptions, e v aluating the impact of adv erse weather on k e y netw ork parameters becomes crucial for resilient telecommunication infrastructure. The study specically tar gets the performance e v aluation of a 4G cellular netw ork under adv erse weather conditions, particularly during the rain y season in tropical climate. This study focuses on the signicant impact of en vironmental f actors on netw ork performance, which is cru- cial for optimizing service quality in adv erse en v i ronmental condition by e xamining KPIs parameters (RSRP , RSRQ, SINR, and RSSI. Existing research on the impact of adv erse we ather condition on 2G, 3G, and 4G netw orks sho w that notable g ap e xists re g arding the impact on 4G netw orks. This s tudy aims to ll this g ap by specically e v aluating 4G netw ork per formance during hea vy rainf all by monitoring and analyzing KPIs (RSRP , RSRQ, SINR, and RSSI) during hea vy rainf all. The approach used in this study contrib utes to a deeper understanding of ho w these metrics int eract under adv erse weather , highlighting SINR as a critical f actor inuencing the NPI. The study identies SINR as signicantly correlated with NPI during the rain y season, stating its import ance for service quality . That adv erse weather af fects the RSSI and SINR which contrib ute to the o v erall netw ork quality . That during adv erse weather conditions that netw ork parameters were af fected. Their result compared the ef fect of weather condition on netw ork parameters during dry and rain y seasons, which suggests decline in netw ork performance in rain y season as drop and block call rates were higher during the rain y season demonstrates that rain-induced impairments can lead to increased call drops, reduced data speeds, and compromised service a v ailability . Their nding is consistent with result of the study depicting where the SINR w as 10dB 50%, the NPI ≥ 2 . 8 . This is a clear indication that in adv erse weather conditions, the SINR is greatly af fected which suggests that signal quality within that period will be dra stically af fected. These nding are not conned to tropical climates; temperate re gions also e xperience similar ef fects, though, to v arying de grees. Corr elation-based assessment of 4G L TE network performance during r ainfall ... (Ngozi C. Eli-Chukwu) Evaluation Warning : The document was created with Spire.PDF for Python.
112 ❒ ISSN: 2502-4752 The impact of rainf all on 4G netw orks, as highlighted in this article, re v eals signicant challenges in maintaining QoS during hea vy rainf all. Rain attenuation, characterized by signal de gradation due to the ab- sorption, scattering, and reection of radio w a v es by raindrops, poses substantial hurdles for netw ork operators. This de gradation af fects KPIs lik e RSRP , RSRQ, SINR, and RSSI, which are essential metrics for assessing the netw ork QoS and user e xperience. Consequently , the study identies a research g ap in the lack of specic optimization strate gies for enhancing SINR under adv erse weather conditions. This g ap emphasizes the need for practical solutions and technological adv ancements to impro v e netw ork performance in such conditions. Comparing the impacts of RSRQ and RSSI on NPI, the study suggests a broader comparati v e analysis across dif ferent weather conditions and netw ork generations. In conclusion, while rainf all poses signicant challenges to cellular netw ork performance, proacti v e measures rooted in technological inno v ation, re gulatory o v ersight, and comprehensi v e netw ork management strate gies are pi v otal in safe guarding the inte grity and reliability of wireless communications, thereby meeting the e v olving demands of global connecti vity in the 21st century . FUNDING INFORMA TION Authors state no funding in v olv ed. A UTHOR CONTRIB UTIONS ST A TEMENT This journal uses the C o nt rib utor Roles T axonomy (CRediT) to recognize indi vidual author contrib u- tions, reduce authorship disputes, and f acilitate collaboration. Name of A uthor C M So V a F o I R D O E V i Su P Fu Ngozi C. Eli-Chukwu ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Uma Uzubi Uma ✓ ✓ ✓ ✓ ✓ ✓ ✓ Handel Emezue ✓ ✓ ✓ ✓ ✓ ✓ ✓ Ogechi Akudo Nw ogu ✓ ✓ ✓ ✓ ✓ Og ah E. Og ah ✓ ✓ ✓ ✓ ✓ Calister N. Ogbonna-Mba ✓ ✓ ✓ ✓ ✓ Samuel I. Ezichi ✓ ✓ ✓ ✓ ✓ C : C onceptualization I : I n v estig ation V i : V i sualization M : M ethodology R : R esources Su : Su pervision So : So ftw are D : D ata Curation P : P roject Administration V a : V a lidation O : Writing - O riginal Draft Fu : Fu nding Acquisition F o : F o rmal Analysis E : Writing - Re vie w & E diting CONFLICT OF INTEREST ST A TEMENT The authors declare that the y ha v e no kno wn competing nancial interests or personal rela tionships that could ha v e appeared to inuence the w ork reported in this paper . D A T A A V AILABILITY The data that support the ndings of this study are a v ailable on request from the corr esponding author , [Uma U.U.] REFERENCES [1] M . T ornatore et al. , “ A surv e y on netw ork resilienc y methodologies ag ainst weather -based disruptions, ” in 2016 8th International W orkshop on Resilient Networks Design and Modeling (RNDM) , 2016, pp. 23–34, doi: 10.1109/RNDM.2016.7608264. [2] N . C. Eli-Chukwu, J. M. Aloh, and C. O. Ezeagwu, “ A systematic re vie w of arti cial intelligence applications in cellular netw orks, ” Engineering , T ec hnolo gy and Applied Science Resear c h , v ol. 9, no. 4, 2019. [3] A. H. A. Salih, A. T . Saeed, and Z. R. Saber , “ A study on the ef fects of cellular mobile netw orks on people in tikrit city based on po wer density measurements and calculations, ” Engineering , T ec hnolo gy and Applied Science Resear c h , v ol. 9, no. 3, 2019. [4] S. Sab u, S. Renimol, D. Abhiram, and B. Premlet, “Ef fect of rainf all on cellular signal strength: A study on the v ariation of rssi at user end of smart-phone during rainf all, ” in 2017 IEEE Re gion 10 Symposium (TENSYMP) , 2017, pp. 1–4. Indonesian J Elec Eng & Comp Sci, V ol. 42, No. 1, April 2026: 105–114 Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752 ❒ 113 [5] A. Markham, N. T rigoni, and S. Ell w ood, “Ef fect of rainf all on link quality in an outdoor forest deplo yment, ” in International Confer ence on W ir eless Information Networks and Systems (WINSYS) , 2010, pp. 1–6. [6] P . K. Sahu, S. R. P al , and A. K. Das, Estimation and Infer ential Statistics . Springer , 2015. [7] N. C. Eli-Chukwu and G. N. Onoh, “Experimental study on the impact of weather conditions on wide code di vision multiple access signals in nigeria, ” Engineering , T ec hnolo gy and Applied Scie nce Resear c h , v ol. 9, pp. 3998–4001, 2019, doi: 10.48084/etasr .2630. [8] A. M. Al-Samman, M. H. Azmi, Y . A. Al-Gumaei, T . Al-Hadhrami, T . Abd. Rahm an, Y . F azea, and A. Al -Mqdashi, “Millime- ter w a v e propag ation measurements and characteristics for 5g system, ” Applied Sciences , v ol. 10, no. 1, p. 335, 2020. [Online]. A v ailable: https://www .mdpi.com/2076-3417/10/1/335 [9] F . Y aghoubi, J. Chen, A. Rostami, and L. W osinska, “Mitig ation of rain impact on micro w a v e backhaul netw orks, ” in 2016 IEEE International Confer ence on Communications W orkshops (ICC) , 2016, pp. 134–139, doi: 10.1109/ICCW .2016.7503777. [10] B. F ong, P . Rapajic, A. F ong, and G. Hong, “Polariz ation of recei v ed signals for wideband wireless communications in a hea vy rainf all re gion, ” IEEE Communications Letter s , v ol. 7, no. 1, pp. 13–14, 2003, doi: 10.1109/LCOMM.2002.806459. [11] F . Beritelli, G. Capizzi, G. LoSciuto, C. Napoli, and F . Scaglione, “Rainf all estimation based on the intensity of the recei v ed signal in a lte/4g mobile terminal by using a probabilistic neural netw ork, ” IEEE Access , v ol. 6, pp. 30 865–30 873, 2018. [12] Q. F an et al. , “Reducing the impact of hando v ers in ground-to-train free space optical communications, ” IEEE T r ansactions on V ehicular T ec hnolo gy , v ol. 67, no. 2, pp. 1292–1301, 2018, doi: 10.1109/TVT .2017.2754960. [13] J. Rak, Resilient Routing in Communication Networks , 1st ed. Springer , 2015. [14] S. P . Thiag arajah, S. Pillay , S. Darmaraju, R. Subramanian, and M. F . M. Fung, “The ef fect of rain attenuation on s-band terrestrial links, ” in 2013 IEEE Symposium on W ir eless T ec hnolo gy and lications (ISWT A) , 2013, pp. 192–197. [15] S.-H. F ang and Y .-H. S. Y ang, “The impact of weather condition on radio-based distance estimat ion: A case study in gsm netw orks with mobile measurements, ” IEEE T r ansactions on V ehicular T ec hnolo gy , v ol. 65, no. 8, pp. 6444–6453, 2016, doi: 10.1109/TVT .2015.2479591. [16] G. Hendrantoro, R. Bultitude, and D. F alconer , “Use of cell-site di v ersity in millimeter -w a v e x ed cellular systems to combat the ef fects of rain attenuation, ” IEEE J ournal on Selected Ar eas in Communications , v ol. 20, no. 3, pp. 602–614, 2002, doi: 10.1109/49.995519. [17] D. P aul, Gaura v , I. Jindal, and A. Sharma, “Performance enhancement of 4g lte netw ork during rain y weather by bit er ror rate (ber) reduction, ” J ournal of Emer ging T ec hnolo gies and Inno vative Resear c h (JETIR) , v ol. 7, no. 12, pp. —, Dec 2020, uGC Appro v ed Journal No: 63975, December 14, 2020. [Online]. A v ailable: htps://www .jetir .or g/papers/JETIR2012123.pdf [18] Y . Xu, K. W u, J. A. Zhang, Z. W ang, B. A. Jaya wickrama, and Y . J. Guo, “Smartphone-based e xperimental analysis of rainf all ef fects on lte signal indicators, ” Sensor s , v ol. 25, no. 2, p. 375, 2025, doi: 10.3390/s25020375. [19] D. Micheli, G. Muratore, A. V annelli, A. Scaloni, M. Sgheiz, and P . Cirella, “Rain ef fect on 4g lte in-car electromagnetic propag ation analyzed through mdt radio data measurement reported by mobile phones, ” IEEE T r ansactions on Antennas and Pr opa gation , v ol. 69, no. 12, pp. 8641–8651, 2021, doi: 10.1109/T AP .2021.3090505. [20] A. L. Sheu, O. Olabisi, A. S. Ade wumi, O. A. Aremu, and I. A. Azeez, “Quantitati v e assessment of tropospheric impair - ments on cellular netw ork reliability using k e y performance indicators, ” SSRN, T echnical Report, 2025. [Online]. A v ailable: https://ssrn.com/abstract=5650689 [21] T . P atra and S. K. Mitra, “Performance analysis of a rain f ading predicted model in tropical areas for 5g communication, ” Pr o gr ess In Electr oma gnetics Resear c h M , v ol. 110, pp. 49–59, 2022. [Online]. A v ailable: https://www .jpier .or g/PIERM22021904.html [22] J. Abdullahi and A. A. Abdulhamid, “Ev aluating 4g netw ork performa nce in north-central nigeria: A dri v e test-based assessment of k e y performance indicators, ” International J ournal of Resear c h and Scientic Inno vation (IJRSI) , v ol. 12, no. 3, pp. 305–323, 2025. [Online]. A v ailable: ttps://rsisinternational.or g/journals/ijrsi/digital-library/v olume-12-issue-3/305-323.pdf [23] S. Sab u, S. Renimol, D. Abhiram, and B. Premlet, “Ef fect of rainf all on cellular signal strength: A study on the v ariation of rssi at user end of smartphone during rainf all, ” in 2017 IEEE Re gion 10 Symposium (TENSYMP) . IEEE, 2017, pp. 1–4. [24] M. S. Ojo, S. Misra, and R. Ade wumi, “Impact of rain attenuation on wireless communication netw orks in tropical re gions: A surv e y , ” IEEE Access , v ol. 6, pp. 58 884–58 894, 2018, doi: 10.2482/haig an.58.894. [25] O. J. Igbek ele, J. T . Zhimw ang, and E. P . Ogherohw o, “Ev aluation of propag ation losses due to rain attenuated signal on terrestrial radio links o v er jos, plateau state nigeria, ” Physical Science International J ournal , v ol. 23, no. 1, pp. 1–8, 2019. BIOGRAPHIES OF A UTHORS Ngozi C. Eli-Chukwu an Associate Professor , Lecturer , and researcher in the Department of Electrical and Electronic Engineering at Ale x Ekwueme Federal Uni v ersity Ndufu-Alik e, Ebon yi State, Nigeria, where she has also serv ed as the Director of Information and Communication T echnol- ogy (ICT). She recei v ed the B.Eng. de gree in Electrical and Electronic Engineering in 2003 and the M.Eng. de gree in Electrical and Electronic Engineering (Communication Systems option) in 2008 from Enugu State Uni v ersity of Science and T echnology , Nigeria. She also obtained a Postgraduate Diploma i n Project Management from the Uni v ersity of Li v erpool, United Kingdom, i n 2015, and the Ph.D. de gree in Electrical and Elect ronic Engineering (Communication Systems option) from ESUT , Nigeria. She is a Council for the Re gulation of Engineering in Nigeria (COREN)–re gistered engineer and a Fello w of the Nigerian Society of Engineers (FNSE). Her research interests include wireless communications, radio w a v e propag ation, 4G/5G netw ork performance analysis, rain atten- uation modeling, and the application of articial intelligence in communication and sensing systems. She has published widely in reputable high-impact journals, and her w orks ha v e recei v ed signicant academic citations. She can be contacted at email: ngozieli@gmail.com. Corr elation-based assessment of 4G L TE network performance during r ainfall ... (Ngozi C. Eli-Chukwu) Evaluation Warning : The document was created with Spire.PDF for Python.
114 ❒ ISSN: 2502-4752 Uma Uzubi Uma recei v ed the B.Eng. and M.Eng. de grees in electrical and electronic engineering from the Federal Uni v ersity of T echnology , Owerri, Nigeria, in 2006, and the Uni v ersity of Port Harcourt, Nigeria, in 2011, respecti v ely . He recei v ed t he Ph.D. de gree in electrical engi- neering, with a specialization in po wer systems and high-v oltage engineering, from the Uni v er sity of Nigeria, Nsukka, Nigeria, in 2021. He is currently a Postdoc toral Researcher with the Department of Computer Sci ence, Uni v ersidade Federal de La vras, Brazil. He is also a Senior Lecturer with the Department of Electrical Engineering, Ale x Ekwueme Federal Uni v ersity Ndufu-Alik e (AE-FUN AI), Nigeria, where he serv ed as the Head of Department from September 2022 to May 2025. His research interests include po wer system protection, smart grids, po wer system planning, and high-v oltage en- gineering. He can be contacted at email:uma.uzubi@funai.edu.ng. Handel Emezue is a Lecturer in the Department of Electrical and Electronic Engi neering at Ale x Ekwueme Federal Uni v ersity , Ndufu-Alik e, Ebon yi State,Nigeria. He holds an MSc de gree from the Uni v ersity of Ibadan and a BEng (Second Class Upper Di vision) in Computer Engineering from Caritas Uni v ersity . His research interests include softw are engineering, embedded systems, intelligent systems, cryptograph y , speech recognition, and smart monitoring systems. He has serv ed as a consultant on national and international ICT projects, including UNICEF-supported initiati v es. He is a re gistered engineer with the Council for the Re gulation of Engineering in Nigeria and a member of IEEE and IAENG. He can be contacted at email: handel.emezue@funai.edu.ng. Ogechi Akudo Nw ogu an recei v ed her de gree in Electrical/Electronics Engineering from Ale x Ekwueme Federal Uni v ersity , Ndufu-Alik e, Nigeria. Her research interests include embed- ded systems, rene w able ener gy systems, and intelligent control. She has been in v olv ed in se v eral academic and research acti vities related to m odern electrical engineering applications. She ca n be contacted at email: ogechi.nw ogu@funai.edu.ng. Ogah E. Oga is a lecturer in the Department of Electrical/Electronics Engineering at Ale x Ekwueme Federal Uni v ersity , Ndufu-Alik e, Nigeria. He obtained his academic qualications in Electrical Engineering and has se v eral years of teaching and research e xperience. His research focuses on po wer systems, smart grid technologies, and ener gy management. Calister N. Ogbonna-Mba is af liated with the Department of Electrical/Electronics Engineering at Ale x Ekwueme Federal Uni v ersity , Ndufu-Alik e, Nigeria. Her research interests in- clude signal processing, communic ation systems, and applied electronics. She acti v ely contrib utes to research and academic de v elopment in her eld. Samuel I. Ezichi is a lecturer in the Department of Electrical and Electronic Engineering at Ale x Ekwueme Federal Uni v ersity Ndufu-Alik e, Ebon yi State Nigeria, where he specializes in engi- neering inno v ation, machine l earning, and wireless communications. His research e xplores cutting- edge technological solutions to contemporary challenges in communication systems and intelligent automation. W ith a strong commitment to interdisciplinary collaboration, Samuel contrib utes to ad- v ancing the eld through scholarly publications and academic mentorship. He can be contacted at email: ezichi.samuel@funia.edu.ng. Indonesian J Elec Eng & Comp Sci, V ol. 42, No. 1, April 2026: 105–114 Evaluation Warning : The document was created with Spire.PDF for Python.