This research work use Dynamic Spectrum Allocation (DSA) embedded in 5G technology and logistic regression as Machine learning tool to alter the reuse factor of assigned telecommunication frequency to a more robust and reliable frequency. The switch from a frequency reuse of 3 (FR3) to a frequency reuse of 7 (FR7) was considered. Historical data generated from a Telecommunications Company in Nigeria was utilised. This data consist of User throughput and the number of user equipment to assess traffic load and congestion from user equipment (UE).The ideal was meant to determine the probability at which a switch or transition between reuse factors will occur. With a resolution threshold of 50 users which was set to commence the shift from reuse factor of 3 when the number of users have reached its threshold of 50 to reuse factor of 7 when the number of users exceeds 50 and above. The objective of this combination is to increase spectrum utilisation efficiency and network flexibility, by optimising spectrum resources in densely populated urban areas. This study explains the mechanisms and also demonstrate how logistic regression with DSA will enhance network capacity of wireless communication systems. The logistic regression model proved effective in making intelligent decisions regarding the transition from reuse factor 3 to reuse factor 7 as it was able to accommodate up to 79 users as against the threshold value of 50 users when running under the reuse factor of 3. With a high accuracy of 91.9% and balanced precision-recall values, it showcased its ability to optimize spectral efficiency.
Read full abstract