Ethernet is a local area network (LAN) technology. This system consists of several rules for connecting several systems to the LAN connection. In addition to the LAN, it is also used in the metropolitan area network (MAN) and the wide area network (WAN) [1] . Its advantages include the ease with which wireless transmission can be set up, and connection-free communication before transmitting data over the network. But for reasons of high cost, people are forced to turn to Wi-fi wireless networks to send and receive data at very low cost [2] [3] . With advances in technology, Ethernet can now cover some ten kilometers using fiber optics. Today, it is becoming a necessity for industries in all communication applications. The cost of using the Internet in Africa is very high. For example, prices for 1 GB in some African countries varied between 2.21, 3.17$ and sometimes 15$ [4] . So, with the purchasing power of the African population, it would be difficult for us to use the Internet connection. To remedy this problem, the government of the Republic of Guinea, our country, has undertaken extensive work since 2010 to improve this system by installing over 2000 km of fiber optics throughout the country; this will revolutionize certain activities in the country and the opportunity to open up new markets, create jobs, speed up the buying and selling process and so on [5] . With all these efforts, this sector will be able to meet the needs of the population in general and Guinea’s education system in particular. Any other device capable of improving connection quality and performance is up for grabs. To this end, we have set up a device capable of specifically testing the performance of the Wi-Fi network at IST Mamou (Guinea). Specific objectives include characterizing usage patterns (frequency, devices, types of users) of Wi-Fi among students and teachers, identifying the main factors contributing to fluctuations in Wi-Fi network performance indicators, highlight the spatial areas and time periods most affected by poor network performance, and make data-driven recommendations to improve network efficiency and user experience.

In order to provide a more comprehensive understanding of internet usage trends and network performance at Mamou Institute of Technology, we conducted additional statistical analyses beyond the initial descriptive statistics.

A Pearson correlation analysis was performed to explore the relationships between the number of weekly connections, device type used, and network performance indicators (download and upload speeds) ( Table 2 ). This allowed us to determine whether user behavior and device preferences have a statistically significant impact on performance outcomes.

Key Findings:

These findings strengthen the understanding of how student behavior intersects with technical infrastructure. They reveal that device type has a notable impact on user experience, and that upload speeds may be better for high-frequency users due to behavioral or technological factors. This deeper insight highlights the need for network upgrades that particularly improve performance for mobile devices, such as additional access points or mobile-optimized bandwidth allocation.

Presentation of network performance mean and standard deviation to IST on October 29, 2024.

The mean and standard deviation of network performance at IST on October 29, 2024. are shown in Table 3 .

By observing Figure 4 , we note that the upload rate of October 25 is higher than the upload rate of October 26 between 9:11 and 10:11 and between 11:20 and 12:30, a sawtooth variation between 10 and 11 o'clock. Regarding the day of the 26th, the transfer rate increases between 9:11 and 9:20, then we see a considerable decrease until 11:11, this would be due to the number of high users in the network, in particular, students and teachers are used to transferring files in this time interval. We note a great variability in the file transfer rate between 11:11 and 12:41. This can be justified by the fact that there is a great variability in the arrival and departure of students in the establishment during all the 2-hour periods corresponding to the duration of a course.

As for the day of October 25, 2024, the flow rate is constant and low between 9:55 a.m. and 10:39 a.m., then we see an exponential growth in the flow rate up to more than 2000 Mbps then a rapid decrease between 10:39 a.m. and 12:11 p.m. and an increase between 12:11 p.m. and 12:30 p.m. ( Figure 5 ) illustrates these variations. On October 26, 2024, there was growth between 9:11 and 9:25, then a decrease between 9:25 with a peak of 1501Mbps, followed by a strong increase between 9:25 and 11:09 and a large variability between 11:12 and 12:50.

Looking at ( Figure 6 ), on October 25, 2024, between 9:41 a.m. and 10:11 a.m., we see a growth in downloading with a peak above 14 Mpbs at 10:00 a.m., then it remains constant between 11:11 a.m. and 11:08 a.m., then it increases between 11:08 a.m. and 12:10 p.m. and finally decreases until 12:30 p.m. However, on October 26, 2024, the download is much slower and constant in the range of 0 to 1.10s between 9:11 a.m. and 12:41 p.m.

Looking at Figure 7 , the day of January 16, 2025, the flow rate decreases from 9:00 to 10:10, then remains constant between 10:10 until 14:05, then increases to reach beyond 2.5 Mbps to reach its peak at 15:00, then decreases until 16:01 finally increases between 16:01 and 17:00. Unlike the day of October 29, 2024, the flow rate undergoes a variation between 9:00 a.m. and 10:30 a.m., then increases between 10:30 a.m. and 11:30 a.m., then decreases until 12:30 p.m., then undergoes a variation between 12:30 p.m. and 4:00 p.m. to reach its peak, then decreases until 4:30 p.m., finally increases until beyond 5:00 p.m.

Looking at Figure 8 , on October 29, 2024, the loading is constant between 9:20 a.m. and 11:30 a.m., then increases to beyond 12Mbps to reach its peak at 12:30 p.m., then decreases until 4:00 p.m. and finally undergoes a variation between 4:00 p.m. until beyond 5:00 p.m. While on January 16, 2025, the load decreases from 9:00 to 10:15, then increases, then remains constant from 10:15 until 14:00, decreases again until 15:05, then increases until 16:00 and finally remains constant until 17:00.

Looking at Figure 9 , on October 29, 2024, the download increases then decreases between 9:00 to 9:50, then increases between 9:50 until 10:30, then decreases between 10:30 until 11:35, then increases between 11:35 until 12:30, then described at 1:00, then increases between 1:00 to 1:30, then decreases between 1:30 until 4:15, then undergoes variability between 4:15 until 5:00 to reach its peak, finally decreases and increases beyond 5:00. During January 16, 2024, the download decreases between 9:00 a.m. and 10:05 a.m., then undergoes a variation between 10:05 a.m. and 11:05 a.m., then increases between 11:05 a.m. and 12:05 p.m. to reach its peak, it decreases between 12:05 p.m. and 3:05 p.m., then increases between 3:05 p.m. and 4:00 p.m., finally decreases until beyond 5:00 p.m. ( Table 4 , Table 5 and Figure 10 , Figure 11 ).

The network performance at IST on January 16, 2025, was measured for data speed, download speed, and upload speed. The mean values and standard deviations for these metrics are shown in Table 6 , providing an overview of both the average performance and its variability.

This study provides valuable insight into the internet usage habits of students and the performance of the Wi-Fi network at the Mamou Institute of Technology (IST). It was observed that students connect to the network an average of 11 times per week, primarily using mobile phones. This confirms the growing importance of mobile access in academic digital usage.

The finding that only 1% of students connect daily, while 30% connect six times per week, suggests that internet use is concentrated on school days—likely due to limited or unreliable access outside the campus. This raises important concerns about digital equity and the need for off-campus access support.

From a technical standpoint, the decrease in average download speed (−22.63) alongside the increase in upload speed (+23.82) may indicate a network configuration that prioritizes upload-intensive activities, such as file submissions, collaborative platforms, or video conferencing. This aligns with the evolving academic practices that require students to engage in more interactive and content-creation-based tasks.

The connection between identified performance issues and proposed solutions becomes clearer when we consider the temporal variation in speeds. The overall 30.58% improvement in data transmission speed between October 2024 and January 2025 might reflect reduced congestion during exam or break periods. This supports the need for adaptive optimization strategies, such as those highlighted [14] , including dynamic bandwidth distribution and smart access point allocation.

Therefore, these performance fluctuations should not be seen merely as technical anomalies but as indicators that call for proactive and targeted network management. The comparative findings from other academic settings further suggest that these challenges are not unique to Mamou IST but reflect a broader need across universities for responsive infrastructure planning.

In light of these interpretations, the study advocates for institutional investments in modernizing network equipment, expanding coverage, and implementing intelligent bandwidth management—especially during peak hours. If adopted within a long-term improvement strategy, such actions could significantly enhance equitable and consistent digital access for all students.

The Wi-Fi network’s performance metrics are important markers for gauging inter net usage. Performance data was gathered for this study using a specially designed programming application over a range of time periods, including two 10-hour sessions and two 20-hour sessions on different days. Mamou IST’s Wi-Fi network performance analysis shed light on the variables affecting network efficiency and stability. Significant differences in network performance were found over the course of the 17-day trial, with Wi-Fi throughput varying by 38.56, download speeds by 29.25, and upload speeds by 31.27%. Peak performance periods were indicated graphically, making it possible to choose the best days and times to use the network. The average throughput at Mamou Institute of Technology performed reasonably well when compared to results from comparable experiments conducted in other nations. Finding times of low throughput and creating plans to improve stability can be made easier with the help of these insights into network performance variations. Together with the gathered database, the application created in this study is a useful resource for the scientific community, enabling additional examination of Wi-Fi network performance in various academic settings across the globe.