Multivariable adaptive handoff spectral model for cognitive radio networks

Abstract

This work presents a multivariable adaptive spectrum handoff model for cognitive radio networks that allows improving secondary user's spectrum mobility from the intelligent selection of spectrum opportunities called MAPMFF. Four algorithms for decision making during a spectrum handoff, were developed with different approaches, fuzzy, feedback statistics, predictive and multichannel, which make up the MAPMFF proposed model. The best spectrum opportunities were selected dynamically based on the following decision criteria, the probability of channel availability, estimated time of channel availability, Signal-to-interference-plusnoise ratio, and bandwidth, which were determined through the modified Delphi method and corresponding weights were calculated through the developed FFAHP algorithm. In order to assess the performance of the developed algorithms, a comparative analysis was made between these algorithms and the most relevant spectrum handoff algorithms in the current literature. Unlike other related papers, benchmarking was validated through a trace of real spectrum occupation data captured in the frequency GSM band, which characterizes the real behavior of primary users. In the validation phase, four assessment scenarios were proposed to 40 Cesar Hernández et al. consider, two kind of applications, real-time and best-effort; two traffic case scenarios, high and low and, ten evaluation metrics, number of handoff, number of failed handoff, bandwidth, delay, throughput, fairness, number of handoff without interference, number of handoff with interference, number of perfect handoff and number of anticipated handoff. The main contribution of this work is the development of an adaptive spectral model (MAPMFF) with excellent performance according to the results of validation with real occupation spectral data in 4 different scenarios, with four types of traffic, and under ten evaluation metrics. Consequently, the impact reached by the MAPMFF model is the improvement in the quality of service parameters for mobile communications. The proposed multi-model selects the best spectrum handoff algorithm, among the four models developed in this work. This is according to the spectrum characteristics and requirements of the secondary user, providing an efficient and effective process of frequency channel selection. Then, the MAPMFF model behaves as an expert system when inferring from data known from a wireless application and spectral occupation the best spectral handoff algorithm. This is in order to analyze and decide which spectral opportunity is the appropriated to the user requirements by an inference engine. The results of the comparative analysis with other spectrum handoff algorithms show that the MAPMFF model provides the best performance in the ten evaluation metrics for the four proposed scenarios. According to the obtained results, the proposed model has a low average handoff rate, a high percentage of average throughput and average bandwidth, and a low level of average delay.