Optimized channel allocation in emerging mobile cellular networks

Abstract

The task of optimizing service quality in wireless networks is a continuous research that requires the design of efficient channel allocation schemes. The problem is how limited channel resources can be maximally utilized, to guarantee seamless communication while maintaining excellent service quality. Whereas, fixed channel allocation (FCA) schemes treat new and handoff calls equally without preference to normally prioritized handoff calls; dynamic channel allocation (DCA) schemes accommodate users mobility in randomly changing network conditions. However, classical Erlang-B models are deficient and do not consider users mobility and dynamically changing traffic of the mobile network environment. A modified Erlang-B dynamic channel allocation (MEB-DCA) scheme is therefore introduced in this paper, for improved network performance. The MEB-DCA algorithm introduces a conditional threshold for handoff request assignment to ensure that communication systems do not unnecessarily prioritize handoff calls at the detriment of new calls. Deriving knowledge from imprecise network data is difficult when developing functional relationships between parameters, requiring advanced modeling techniques with cognitive experience. Soft computing techniques have been shown to handle this challenge given its ability to represent precisely, both data and expert knowledge. An adaptive neuro-fuzzy inference system-based dynamic channel allocation (ANFIS-DCA) framework was proposed to automate the learning of communication parameters for optimized channel allocation decisions. Network parameters considered were received signal strength indication impacted by user mobility, number of guard and general channels, carried traffic, and handoff blocking threshold. The performance of the proposed ANFIS-DCA model was found to outsmart the static FCA and back propagation neural network-based DCA (NN-DCA) schemes using mean square error and root mean square error as performance measures. Our approach can be effectively deployed to improve channel allocation, resource utilization, network capacity, and satisfy users experience.