Research On Key Technologies In Broadband Cooperative Relaying System Based On OFDM

As a novel wireless communication paradigm, the unique idea behind cooperative relaying is to make distributed user nodes with different physical locations acting as relays to help source node send signals. By exploiting the concept of virtual MIMO, cooperation among users can provide the extra spatial diversity, which is particularly appropriate for the small sized wireless terminals due to their difficulty in equipping multiple antennas. Besides, cooperation also has advantages in system capacity, spectral efficiency and coverage area. On the other hand, for the high data rate broadband communication, OFDM is the promising technique to resist the frequency-selective fading with advantages of low implementing complexity and flexibility in band resource allocation. It has wide applications in current and future mobile telecommunication systems.Combining cooperative relaying with OFDM can exploit both advantages and has important application prospects. However, it also poses a couple of technical obstacles significantly different from the narrow-band cooperative schemes. Therefore, how to use cooperation effectively in OFDM system becomes increasingly important. The dissertation is thus to investigate several key issues on this subject such as channel estimation with training design, equalization technique to resist multiple carrier frequency offsets (CFO), diversity performance of broadband cooperative OFDM, low-complexity relay selection (RS) schemes and beamforming (BF) using feedback of channel state information (CSI). The main contributions are summarized as follows:To begin with, channel estimation and training design are investigated for using multiple amplify-and-forward relays in frequency-selective channels. To make the channel being identifiable, we perform cyclic convolution filtering (CCF) at relays before forwarding and then derive a suboptimal design of training sequence and CCFs. Furthermore, it is shown that this training design is actually asymptotically optimal as the increasing of the SNR of source-to-relay links. To further improve the performance, the closed expression of power allocation between source and relays is derived. Simulation results show that the proposed scheme has high estimation precision and can boost the system performance significantly as compared with non-optimized trainings.Second, based on the cooperative OFDM employing orthogonal space-time block coding, the equalization methods are investigated to resist the signal interference caused by multiple CFOs. Since the design requirement varies with the CFO condition, we first propose an iterative algorithm to achieve the zero forcing equalization. whose main procedures can be accomplished by FFT/IFFT operations. Due to the very low complexity. this scheme is particularly suitable for the scenario where the difference of CFOs (DCFO) is small as well as the interference. Moreover, to resist the severe interference more efficiently when DCFO is large, another equalization method is proposed where the interference cancellation is first performed among subchannels and then the interference among all the signals is canceled parallelly. It can improve the error performance dramatically as compared to some existing methods. Based on our design, the matrix inversion required by equalization can be computed recursively. As a result, the computation complexity is reduced significantly and thus the availability in practice is ensured.Next, the diversity performance of cooperative OFDM with fixed gain amplify-and-forward (FG-AF) relays is investigated. Earlier studies have shown that the diversity performance for narrow-band FG-AF relaying should be characterized by the diversity gain function (DGF) with form of ((logρ/ρ)X), whereρstands for the signal-to-noise ratio. Inspired by this, based on the OFDM transmission employing linear constellation precoding, we analyze the diversity performance as well as its relationship with the precoding group size J for using FG-AF relaying over broadband channels. Suppose the number of multipath components for the source-to-relay channel is L1 and that for the relay-to-destination channel is L2. By analyzing the pairwise error probability (PEP), we show that if L1≠L2, the optimal DGF is preciselyρ-L with L=min{L1,L2}, which can be achieved sa long as choosing J≥L+1, whereas only a DGF of (logρ/ρ)J is obtainde by precoding with size J≤L. In the case of L1=L2, it is shown that the best achievable DGF is logρ/ρL, which can be guaranteed by choosing J≥2L-1. Since the PEPs of the distributed space-frequency coding (DSFC) scheme and the single-relay precoding system are closely linked, these above results can be easily extended to the multi-relay case.Moreover, to reduce the high complexity of diversity-oriented multi-relay cooperative OFDM, the relay selection (RS) methods for precoded decode-and-forward relaying are studied. Specifically, we propose two different RS schemes. One is to select the best relays for individual precoding groups, and the other is to choose a single relay for the entire precoding groups. By evaluating the group error performance, RS criterion for each scheme is obtained. From a further analysis, it is shown that with the proposed criterion, both schemes can achieve the full cooperative and frequency diversity. Numerical results also show that the former scheme can approach the performance of DSFC scheme while keeping a much lower en/decoding complexity, whereas the latter one has some inevitable performance loss as a result of easy realization and less communication overhead. Besides, to further reduce the decoding complexity, an improved sphere decoding algorithm is proposed. By accelerating the expansion of the bottom node in searching structure, the proposed algorithm has much lower computation complexity than conventional schemes, only at a price of slightly increasing the error probability.Finally, to improve the performance of close-loop system, we investigate the cooperative OFDM beamforming with full CSI. Based on the separate relay power constrains, the optimization algorithms for relay BF as well as the joint relay BF and source power allocation are derived, aiming at maximizing the sum capacity. By using the Lagrange dual theory, we show that at the dual domain, both problems can be solved by iterative computing a series of single-variable subproblems, which are solvable by line search methods, with guarantee of convergence to global dual optima. Therefore, both proposed algorithms have low complexities. These algorithms are asymptotically optimal as the number of subcarriers goes to infinity, and for the finite number of subcarriers, we show at least the feasible suboptimal solution of primal problem can be found necessarily. Moreover, it is learned from the special forms of optimal BF weights that the proposed methods are suitable for realization in a distributed manner and then the communication overhead is reduced. To further boost the performance, we also embed the subcarrier paring into our BF framework and derive a practical optimization algorithm. Simulation results show that our proposed schemes can improve the capacity and reduce the outage probability significantly.