Title: Forecasting of HOSR for Different Mobile Carriers in Kano Using Conventional and Intelligent Techniques

Issue Number: Vol. 9, No. 1
Year of Publication: March - 2019
Page Numbers: 1-10
Authors: S. B. Abdullahi, M. S. Gaya
Journal Name: International Journal of New Computer Architectures and their Applications (IJNCAA)
- Hong Kong
DOI:  http://dx.doi.org/10.17781/P002599


Handover is the key concept to achieve mobility and its success rate determine the subscriber satisfaction. It is a critical process and if performed incorrectly can result in the loss of calls. Therefore, reliable and accurate prediction of Handover Success Rate (HOSR) is very essential. The available models did not well capture the impact of selected inputs dataset of HOSR. The comparison between the conventional technique-Hammerstein-Wiener (HW) and intelligent technique-Adaptive Neuro-fuzzy Inference System models in forecasting HOSR are considered. Seven months log files were generated from four active GSM-1800 networks in Kano Metropolis with the aid of TEMS Pocket. The collected data were embedded in the construction of the trained networks. The results depicted that Conventional technique is superior to the intelligent technique for predicting HOSR. The performance of the linguistic models was tested using MAPE, where ANFIS achieved 0.3358, 0.2834, 0.0092, 0.9519: and HW achieved 0.0000, 0.0683, 0.1810, 0.0568 and with RMSE, ANFIS achieved 50.5677, 39.4984, 6.8271, 139.6244: and HW achieved 0.0000, 9.4974, 134.4543, 45.6185 for P, Q, R and S respectively. Thus, the Conventional technique would be suitable in forecasting HOSR in Kano or any environment with similar network deployment