Research On Iterative Reweighted Channel Estimation And Hybrid Precoding Algorithms For Millimeter Wave MIMO Systems

With the rapid development of Internet technology,the information generated in today's network environment is often huge and complex,people are swallowed up by a large amount of information,in the face of these dazzling massive amounts of data,users have no direct and convenient access to the valid information they need.How to efficiently provide people with the information they need is the key to today's Internet environment.The most effective method for this information overload problem is the recommendation system.It analyzes people's potential interests according to their historical information to help users discover the information that they may need it.The information saves users time to find information.Although the recommendation system can alleviate the information overload problem,there are still some problems to be solved.Because the recommendation process is based on existing information of users,it faces the problem of poor recommendation accuracy in the face of absent user information.On the other hand,because the recommendation system needs to grasp user information,it will inevitably have threats to hamper user privacy.The causes of these two problems are contradictory.How to solve these two problems at the same time is a subject worthy of being explored in depth.This paper first expounds the research background of the recommendation system,points out the research significance of this paper,summarizes the research status at home and abroad,and summarizes the advantages and disadvantages of the commonly used recommendation models and privacy protection methods.Then,the working principle of the popular recommendation algorithm and the basic definition of differential privacy are detailed,and the corresponding solutions are proposed based on the combination of auxiliary information and differential privacy.The specific research contents of this paper are as follows:(1)To improve the ability to recommend cold start items and protect user privacy,this paper proposed a matrix factorization recommendation model combining differential privacy and tag information.Firstly,the model added the tag information to the process of calculating item similarity,then integrated it into the recommendation model to improve the recommendation accuracy.Finally,this paper solved the model optimal value by the stochastic gradient descent method.For protecting users from privacy threats,the proposed approach divided Laplace noise into two parts,which are added to the process of item similarity and gradient solution respectively,so that the whole recommendation process satisfied the differential privacy,and analyzed the validity of the algorithm on a real data set.Experimental results show that the proposed method has high recommendation accuracy while protecting users' privacy.(2)To improve the recommendation capability for cold-start users,considering the social relationship of users,a differential privacy indirect social relationship recommendation algorithm is proposed.Firstly,the social relationship matrix is populated by the user indirect relationship.The social relationship matrix is added to the decomposition model of the user rating matrix.Then synchronously decompose rating matrix and the social relationship matrix to share the user feature vectors.The user eigenvectors in the user rating matrix and the social relationship matrix are assumed to be same.To protect user privacy,Laplace noise is added to the user rating and user social relationship respectively to protect the user's sensitive rating and social relationship to reach the differential privacy objective-perturbation technique.Experimental results show that the proposed algorithm achieves good performance.