Analysis of Spectrum Occupancy Prediction Results for Maitama Abuja

Johnson Adegbenga Ajiboye¹, Mary Adebola Ajiboye², Babatunde Araoye Adegboye³, Daniel Jesupamilerin Ajiboye⁴, Jonathan Gana Kolo⁵ and Abiodun Musa Aibinu⁶, ^{1, 3, 5}Department of Electrical and Electronic Engineering, Federal University of Technology, Minna, Nigeria, ²Department of ICT, Abuja Electricity Distribution Company, Niger Regional Office, Minna, Nigeria, ⁴Department of Computer Engineering, Federal University of Technology, Minna, Nigeria, ^{5, 6}Department of Mechatronics Engineering, Federal University of Technology, Minna, Nigeria

ABSTRACT

This research analyzes spectrum occupancy in Maitama, Abuja, using Artificial Neural Networks (ANN) to predict usage across frequency bands from 30 MHz to 300 MHz. Predicted spectrum occupancy was compared with actual measurements to assess accuracy. Results for different bands show that prediction errors were generally low, often below 1.5%. The 30-47 MHz band had an average error of 8.7 x 10-2%, peaking at 1.12%. For the 47.05-68 MHz band, the average error was 10.61 x 10-2%, with a maximum of 2.18%. In the 68.05-74.8 MHz band, the average error was 3.99 x 10-2%, peaking at 23.24 x 10-1%. The 74.85-87.45 MHz band showed the lowest average error of 0.99 x 10-2%, peaking at 17.38 x 10-1%. Other bands had similar low error rates, demonstrating the high accuracy of ANN in predicting spectrum occupancy. The highest overall error was 10.61 x 10-2% in the 47.05-68 MHz band, while the lowest was 0.99 x 10-2% in the 74.85-87.45 MHz band.

Keywords

ANN, Prediction, PSD, dBm, VHF.