Semi Blind Channel Estimation In Cloud Radio Access Networks

The rapid development of mobile Internet and Internet of things has a sharp rise in wireless data transmission, which puts forword higher requirement in the performance of wireless communication network. In order to solve the current demand, the cloud radio access networks(C-RANs) consist of centralized base-band pool processing, co-operative radio and real-time cloud infrastructures are proposed. Since the channel property directly determines the gain of concentration signal processing,the method to improve the channel estimation performance in C-RANs is studied in this thesis. The channels in C-RANs are composed of wireless radio access links and fronthaul links. In this thesis, a semi blind channel estimation method is presented, in which data signal is used to improve the performance of training-based channel estimation. Combining with the cooperation characteristics of C-RANs, game theory is utilized to optimize the semi blind channel estimation method in C-RANs with non-ideal wireless fronthaul links.Firstly, in order to reduce the training overhead of channel estimation and improve the transmission rate of the system, data signal is used to improve the estimation accuracy together with training signal in semi blind channel estimation method in view of the ideal fronthaul link C-RANs. This kind of method can improve the estimation performance without increasing the training overhead, which proposes a solution to deal with the contradiction between the estimation accuracy and the cost of training overhead.Secondly, in C-RANs with non-ideal wireless fronthaul links,taking the cooperation characteristic of the remote radio heads (RRHs)into consideration, we presented a joint optimization algorithm of semi blind channel estimation and cluster formation, which balances the channel estimation accuracy and transmission rate. In this algorithm,RRHs act as players in the coalition formation game. By considering both estimation accuracy and transmission rate, the optimal cooperation form is obtained. Simulation results show that the semi blind channel estimation with the optimal cluster form can obtain 1.01 bit/s and 2.6bit/s gains comparing with grand cooperation and non-cooperation,respectively.Thirdly, in the proposed joint optimization algorithm of C-RANs with non-ideal wireless fronthaul links, instantaneous channel state information (CSI) is needed in each iteration of cluster formation, the complexity of the algorithm is high. In order to solve this problem, a low complexity joint optimization algorithm of semi blind channel estimation and clustering formation is presented. By computing the expectation of the utility function, the instantaneous CSI is removed from the utility function. Thus, semi blind channel estimation and cluster formation are decoupled, which reduces the complexity of the algorithm from O[2M(M3+D2 +L2DlogD+LsK2)] to O(2M+M3+D2 L2DlogD+LsK2).Simulation results show the convergence, estimation performance and data transmission performance of the low complexity joint optimization algorithm.