Research On Signal Detection And Parameter Estimation In Mobile Communications Based On Multi-Way Arrays

Blind signal processing technology, which includes the total-blind and semi-blind signal processing technology, can realize the signal detection and paramater estimation without or with only little channel state information (CSI), which saves the valuable bandwidth resources. The multi-way array is a matrix model which can be used to realize the blind signal detection. Compared with two-way arrays, multi-way arrays have the the property in the low rank decomposition. Exploiting this property can get the uniqueness of the composing paramaters, and using the appropriative iterative algorithm can get global optimal solution, which make this technology a hot spot in signal processing area.Under the support of National Science Foundation of China (60872149) and HuaWei Foundation, the signal detection and parameter estimation technologies in Multiple Input Multiple Output (MIMO) and Orthogonal Frequency Division Multiple Access (OFDMA) based on multi-way arrays are investigated. By combining the idea of low rank decomposition of multi-way arrays with mobile communication systems and signal structural feature, novel signal processing algorithms are proposed. Performances of these algorithms are analyzed theoretically and then are compared with other algorithms. The main contributions of this dissertation are as follows:(1) In mobile communication systems, the received signals can be written as a multi-way array including the transmitted signals and other information, in which way the transmitted signals can be blindly detected. In Multi-Hop Alamouti Amplify and Forward (AAF-MH) cooperation system, the received signals only contain transmitted signals and channel information. In order to construct a Parallel Factor (PARAFAC) model including three loading matrices, the third matrix called as linear matrix is introduced under the condition of invariable signal at the receiver. Therefore, a PARAFAC blind signal detection algorithm is proposed. This algorithm can be used not only in the AAF-MH cooperation system, but also in the two-user Alamouti cooperation system with similar structure. In light of the characteristics of different constant modulus iteration algorithm, the initialization work is divided into two steps. Compared with the Constant Modulus Algorithm (CMA), this algorithm can realize the uniqueness of parameters decomposition under the condition of the global convergence, and has advantages in terms of detection performance.(2) In multi-user relays cooperation systems, the co-channel interference exists when each user uses the same resource block. A semi-blind signal detection algorithm in cooperation systems with co-channel interference is proposed. The signals in receiver include the transmitted signals, the channel information from the Mobile Stations to Relay Stations (MS-RS) and from the Relay Stations to Basestation (RS-BS), and the amplying matrix in relays, which can be written as a PARAFAC model with four matrices. Furthermore, a few training sequences are introduced and simplify the received signals to be a new PARAFAC model including the two hop channel matrices (MS-RS, RS-BS) and the amplying matrix, in which way each hop channel information can be estimated. The proformance of this algorithm is better than the two-stage training (TST) semi-blind signal detection algorithm.Existing works show that in the use of array antenna Amplify and Forward (AF) relay cooperation system, the received signals can be written as a PARAFAC model constructed by transmitted signals matrix, array response matrix and the equivalent channel matrix. The transmitted signals can be recovered by the decomposition uniqueness of this model. However, the noise is amplified as colored noise, and the influence of the colored noise is not considered in this algorithm. For this problem, an improved PARAFAC blind signal detection algorithm by means of constructing linear weighting matrix is proposed in this paper, improving the estimation results. Compared to the estimation results only using PARAFAC, this algorithm has better performance at the expense of the complexity. Meanwhile, this algorithm can be widely used in the mobile communication systems with colored noise, which shows the good expandability of this algorithm.(3) In multi-user MIMO uplinks, the interference between users can be reduced or eliminated using precoding technology, which improves the performance of communication systems. A blind signal detection algorithm in multi-user MIMO uplinks based on PARAFAC model is proposed in this paper. Doing the operation of covariation to the received signals, a PARAFAC model including the equivalent channel matrix, the conjugate of equivalent channel matrix and signal matrix, is constructed. Then, the equivalent channel matrix can be estimated by Tilinear Alternating Least Square (TALS) or Joint Diagonalization algorithm. Finally, the signal matrix can be obtained according to the estimation of the equivalent channel matrix. The feasible conditions and computation complexity of the two estimation algorithms are analysed, which shows the advantages of proposed algorithm. Theoretical analysis and simulation results show that the proposed algorithms have better performance in BER and NMSE, and remain valid in small data blocks.(4) In OFDMA uplnks, all the subcarriers can be expressed by an Inverse Discrete Fourier Transform (IDFT) matrix. The subcarriers used are allocated to several users, which can be expressed by a product of IDFT matrix and subcarrier selected matrix. According to this point, a novel blind signal detection algorithm in OFDMA uplinks based on Parallel Profiles With Linear Dependencies (PARALIND) model is proposed. In this algorithm, the received signals can be constructed as a PARALIND model including transmitted signals, channel and IDFT matrices. The transmitted signals can be recovered by iteration. The decomposition uniqueness of this model is demonstrated by derivation. The performance of this algorithm has small difference compared with the semi-blind signal detection algorithm, without pilot sequence, which shows the advantages of this algorithm.