The Algorithm Research On Wireless Channel Allocation And Transmission Power Control In Cognitive Radio

With the rapid development of wireless equipment and related applications, traditional wireless network access, transmission protocol has been unable to adapt to this trend. Among them, the main challenges include the following aspects: 1. the wireless spectrums are almost exhausted; 2. the utilization of allocated wireless spectrums are extremely low; 3. the Quality of Service(Qo S) of wireless communication needs to be improved, especially in busy areas of wireless network.The Cognitive Radio Networks(CRNs) is an effective method to solve the above problems, and the main idea of CRNs is to use intelligent algorithms to solve the problems of access and transmission in wireless network. The advantage of CRNs is that the wireless devices can access different channels in an intelligent way, and this way efficiently improves the Qo S of wireless communication.To solve the problems of multi-channel access and transmission power control in CRNs, the existing algorithms include Q-Learning algorithm, Genetic Algorithm, Game Algorithm, Dynamic Bayesian Network Algorithm and so on. Although these algorithms have a good effect in some cases, there are still many defects: 1. slow convergence; 2. low processing accuracy for the decisionmaking problems which contain float numbers as parameters; 3. poor environmental adaptability; 4. small range of controllable transmission power.Based on this situation, this paper mainly uses the Partially Observable Markov Decision Process(POMDP) and Monte Carlo Value Iterative(MCVI) to solve the problems of channel allocation and power control in wirel ess network. The POMDP is one of the best models in partially observable wireless environments to forecast the interactions between intelligent agents and wireless environment, and the MCVI is effective algorithm to solve the POMDP models which has float numbers as perceived power values, but the MCVI also has a low convergence rate according to the experimental analysis. To solve the defect of MCVI, this paper optimizes the MCVI by using an improved Non-Dominated Sorting Genetic Algorithm(NSGA2), and the optimized algorithm is named Fast MCVI. According to the tests using a classical POMDP's problem, it proves that the convergence rate and decision-making speed of Fast MCVI are both better than original MCVI.Compared with other algorithms, using the Fast MCVI to solve the CRNs problems provides following advantages: 1. accelerating the decision-making speed; 2. improving the credibility of decisions by using float numbers as parameters directly; 3. extending the maximum number of supported wireless nodes and power control range. By simulating the Fast MCVI and other algorithms on Exata, it is proved that the Fast MCVI is better than other algorithms in reducing the wireless transmission delay and increasing the rate of channel multiplexing.