Study On Application Of Hidden Markov Model In Dynamic Spectrum Sharing System

Cognitive radio (CR) is an emerging technology for sensing and dynamic access ofspectrum in mobile radio environments. It aims at dynamically allocating unusedbandwidth among unlicensed secondary users without causing harmful interference tothe primary users. How to avoid the interference to primary users? Secondary usersneed sensing the primary users signal, and then estimate the primary user access modelof channel utilization. Though spectrum measurements, the primary user channel accessmode appears Markov process. The secondary users spectrum sensing obeys a hiddenMarkov model (HMM). Through the research on the spectrum sensing of the hiddenMarkov model, this paper proposes an efficient detection algorithm. Further study inprimary users Hidden Markov state transfer model, we proposed a gradient algorithm toestimate the primary user state transfer parameters.This paper is divided into six chapters:In the first chapter, we mainly introduced the research background and thestructure of this article.In the second chapter, we mainly introduces hidden Markov model, three kinds ofclassic Markov model solutions, and briefly introduces the cooperative awarenesstechnology.In the third chapter, through spectrum measurement on GSM band of powerspectral density (PSD)，by measuring data, training the proposed hypothesis model.Finally, using the simulation data, determine the training system model whether correct,and presents a simple detection algorithm.In the fourth chapter, in order to improve the second user’s spectrum sensingefficiency in hidden Markov model, we proposed a sequence detection algorithm forspectrum sensing. This algorithm can assign the different Bayesian cost factor for falsealarm and missing. Through the comparison of several algorithms, Weightedforward–backward proposed in this paper is optimal in minimizing the detection risk. Inorder to further enhance the efficiency of the algorithm, the complete forward algorithmand the complete forward partial backward algorithm are introduced and theirperformances are compared as well.In the fifth chapter, primary users traffc pattern changes over both time andfrequency according to upper layer events, so we design a dynamically to estimate theprimary user state transfer parameters algorithm. Algorithm based on sensing sub-framediscontinuous setting, using the gradient algorithm to find the primary user parameters.However, a single user in low signal-to-noise ratio (SNR) environment cannot estimate parameter correct. We propose a cooperative sensing algorithm to estimate the PUparameters.In the last chapter, it is a summary of this work, and proposed the further researchdirections.