ABUAD Journal of Engineering Research and Development (AJERD) ISSN (online): 2645-2685; ISSN (print): 2756-6811 Volume 7, Issue 2, 425-435 https://doi.org/10.53982/ajerd 425 Three-State Hidden Markov Model for Spectrum Prediction in Cognitive Radio Networks Emmanuel Oluwatosin RABIU 1 , Damilare Oluwole AKANDE 1 , Zachaeus Kayode ADEYEMO 1 , Isaac Akinwale AKANBI 2 , Oluwole Oladele OBANISOLA 3 1 Electronic and Electrical Engineering Department, Ladoke Akintola University of Technology, Ogbomosho, Nigeria oluwatosin.rabiu@gmail.com/doakande@lautech.edu.ng/zkadeyemo@lautech.edu.ng 2 Nigerian Communications Commission, Abuja, Nigeria engrakanbiia@gmail.com 3 Department of Electrical and Electronic Engineering, Ajayi Crowther University, Oyo, Nigeria oo.obanisola@acu.edu.ng Corresponding Author: doakande@lautech.edu.ng, +2348066133011 Date Submitted: 17/07/2024 Date Accepted: 08/10/2024 Date Published: 13/10/2024 Abstract: The exponential growth and proliferation of wireless devices for different wireless applications have led to the emergence of cognitive radio network (CRN) for optimal utilization of scarce spectrum resources. However, these resources have grossly been under- utilized due to the inaccurate spectrum predictions. Existing spectrum occupancy and prediction techniques which rely on 2-state hidden Markov model (HMM) results in false alarm or missed detection caused by noisy or incomplete observable effects. In this paper, a 3- state HMM spectrum occupancy and prediction technique in CRNs is proposed. The transmission, emission and initial state probabilities of the proposed 3-state HMM parameters were derived based on the three canonical problems associated with HMM. The evaluation, decoding and learning problems were solved using Forward algorithm, Viterbi algorithm and the Baum-Welch algorithm, respectively. The performance of the proposed 3-state HMM spectrum prediction technique was evaluated using prediction accuracy, probability of detection and spectrum utilization efficiency. The simulation results obtained revealed that the 3-state HMM outperformed the 2-state HMM spectrum prediction technique by 24.1% in prediction accuracy. Keywords: Cognitive Radio Network, 3-state HMM, Spectrum Prediction, Prediction Accuracy, Probability of Detection 1. INTRODUCTION A remarkable success recorded over the decade by the wireless communication research community is the tremendous and evolutionary transformation from the first generation (1G) cellular telephone systems to the fifth generation (5G) network [1-3]. Spectrum is an important component of wireless communication and has been reported to be scarce and under-utilized because of different emerging wireless applications. To subdue the challenge of spectrum scarcity, cognitive radio network (CRN) was introduced by Mitola in [4] as one of the important components of the 5G network. The technology helps to increase spectral efficiency as secondary users (SUs) or cognitive radio (CR) users are allowed to access spectrum holes left unused by licensed or primary users (PUs) [5,6]. In order to avoid undue interference to licensed users, there is need for a dynamic spectrum access (DSA) by cognitive radio nodes which require algorithms and protocols for rapid spectrum sensing accuracy, efficient coordination and beneficial cooperation [7-9]. Spectrum prediction is a robust technique complimentary to spectrum sensing for obtaining the relevant information as regards spectral evolution. It also provides identification of spectrum holes as a band of frequencies meant for a certain PU but which are found vacant at some points in time and specific geographic locations [10]. Spectrum sensing involves the determination of the spectrum state in a passive manner using various signal detection methods. Spectrum prediction on the other hand detects the state of radio spectrum from already known spectrum occupancy statistics by effectively utilizing the existing inherent correlations in a proactive manner [11-13]. Spectrum prediction involved different techniques ranging from pure lookup table (LUT) to advanced techniques such as evolutionary algorithms, artificial neural network (ANN), hidden Markov model (HMM) and machine learning (ML) among others. Generally, all spectrum prediction techniques have been categorized into regression analysis-based, machine learning-based and Markov model-based [14-17]. Some of the existing contributions are discussed therein. In the work presented in [18], the design of a spectrum prediction technique using a neural network based and hidden Markov model which is a combination of a regression analysis, and a Markov model was performed. However, the work considered only two physical channel statuses which are