Research On Mechanism Of Traffic Prediction And Data Transmission Based On Artificial Intelligence Algorithm In Cognitive Wireless Network

The cognitive wireless networks can in a certain extent solve the problem of lack of wireless resources,but more is to give the intelligent and adaptive ability to the network entities,at the same time,provides the further technical support for the intelligent management of the wireless network.The entities in the cognitive network can intelligently perceive the external environment information,and make the strategy accordingly.With the increasing of complexity and heterogeneousness of the network environment,the behavior of network entities brings more and more data traffic.Therefore,the intelligent management to the data in the network can effectively improve the performance of the network.So,the research on network data traffic prediction and transmission scheduling mechanism is the important aspects of future intelligent network.Based on this,the main research contents of this paper are as follows:Firstly,the problem of how to improve the prediction accuracy of network data traffic is studied.Aiming at the problem of data traffic prediction in multiple channels network,a traffic prediction model based on stacked auto-encoders deep learning method is proposed,which transforms the precision accuracy problem of traffic prediction into the prediction model optimization problem.Compared with the traditional shallow learning methods,the deep learning method can obtain the deep nonlinear relationship implied in the network data flow.At the same time,the model transforms the single target prediction into multiple targets prediction.Besides,the LBFGS algorithm is introduced to optimize the parameters of the prediction model to improve the prediction accuracy.By simulation experiments and comparing with other algorithms,the results show that the proposed scheme has a smaller prediction relative error.Secondly,the problem of data transmission scheduling in cognitive networks is studied.In the case of one relay node and multiple users in the system,the data forwarding process of relay node is modeled as a Markov decision process model.A deep Q learning based data transmission scheduling scheme is proposed.The Q learning algorithm is adopted to learn and explore the system states transition information in the case of unknown system states transition probability to obtain the approximate optimal strategy of the schedule node.Aiming at the problem of the curse of dimensionality caused by the large amount of computation and storage space in Q learning process in the situation with a large scale,the deep learning method is introduced to establish the mapping relation between state and behavior to accelerate the solving of the problem.The simulation results show that the proposed scheme can approach to the optimal strategy based on the strategy iteration in terms of power consumption,throughput,packets loss rate.And the proposed scheme has a lower complexity,which can solve the problem of the curse of dimensionality.