Search

This paper considers the finite-order filter design problem for the source and multiple full-duplex (FD) relays in a cooperative communication system under frequency-selective fading channels. The goal is to optimize the source and relay filters such that the end-to-end signal-to-interference-plus-noise ratio (SINR) of minimum mean-squared error decision-feedback equalizer (MMSE-DFE) can be maximized. The resultant design problem is very difficult since we need to deal with self interference (SI), inter-relay interference (IRI), and inter-symbol interference (ISI) at the same time. Novel designs are then proposed to overcome the difficulty in this work. Transforming the signals into the frequency domain and using some optimization techniques, we first theoretically derive the power spectrum of the source filter and the spectrums of the relay filters. Then, the finite-order design is developed to approach the derived spectrums. Based on the weighted least-square (WLS) criterion, the Steiglitz-McBride method is exploited to obtain the filter coefficients in finite lengths. Numerical results demonstrated that complete removal of SI may not be a good strategy; preserving a certain amount of SI at each relay instead provides better SINR performance. [ABSTRACT FROM AUTHOR]

Full-duplex (FD) multiple-input multiple-output relaying has been considered an effective scheme to increase the spectral efficiency for wireless communications. As known, the main problem for the FD system is the cancellation of loop interference (LI). In this paper, we propose using the joint source/relay precoding to reduce the influence of LI. Therein, linear precoders are used at the source and relay, while the minimum mean-squared-error receiver is adopted at the destination. The joint precoder design is complicated when spatial multiplexing is exploited for signal transmission. To solve the problem, we propose an iterative method in which the original problem is split into two subproblems. With some matrix properties, we then show that each subproblem can be formulated as a convex optimization. Finally, a closed-form solution can be obtained with the Karush–Kuhn–Tucker conditions. Using a mean-squared-error upper bound, we also propose a low-complexity method to reduce the computational complexity. The proposed precoders have closed-form expressions, which is a great advantage in real-world implementation. Simulation results show that the proposed methods significantly outperform existing ones. [ABSTRACT FROM AUTHOR]

Tomlinson–Harashima precoded (THP) multiple-input–multiple-output (MIMO) systems are developed to cancel the intersymbol interference (ISI) at the transmitter. However, the THP at least requires perfect channel direction information (CDI) to completely mitigate the interference at the transmitter, which is not practical in the real world. The limited feedback systems are then practically developed to convey the quantized CDI to the transmitter. Using the quantized CDI to conduct the THP, an unavoidable interference degrades the performance at the receiver. In this paper, we investigate the performance degradation resulted from the random vector quantization (RVQ), which is a kind of CDI-based limited feedback mechanism. Here, we measure the performance degradation as the excess mean-squared error (EMSE). Since the formulation of the EMSE is a complicated function of the quantization error, we proposed using the second-order Taylor expansion to derive an analytical result. Simulation results have also verified our proposed analysis, showing that the derived analysis can match the actual EMSE. [ABSTRACT FROM PUBLISHER]

In this paper, we study the codebook size design strategies in the multiple-input–multiple-output (MIMO) broadcast channel (BC) systems. We adopt the nonlinear Tomlinson–Harashima precoder (THP), which is designed by the imperfect channel direction information (CDI) acquired from the random vector quantization (RVQ) feedback mechanism. Under the limitation of the total feedback rate, we flexibly allocate the size of each user's codebook so that the overall channel capacity is maximized. Since the design problem is not convex, directly finding out the optimum allocation of codebook size is not attainable. However, by sophisticatedly using the discrete majorization theory, we can derive the codebook size allocation strategies for the high- and the low-signal-to-noise-ratio (SNR) regions. Interestingly, the results show that the equal size codebook is preferable in the low-SNR region, whereas the codebook allocates the whole available codebook size to a single user in the high-SNR region. Simulations certify the theoretical result of our design. [ABSTRACT FROM PUBLISHER]

In this paper, we propose cognitive overlay transceiver designs, where a primary transceiver pair and a secondary transceiver pair coexist in a network, and the primary user (PU) allows the secondary user (SU) to concurrently transmit its signals at the price of reducing the power of the PU's signal relayed by cooperative amplify-and-forward (AF). Since the considered transceiver design is mainly to devise the precoders both for the PU and the SU at the secondary transmitter (ST), the channel state information (CSI) has to be known at the ST. We therefore consider the limited feedback scheme with random vector quantization (RVQ), where the ST can only know the quantized channel direction information (CDI). Considering the statistics of the CSI quantization error and the linear minimum mean square error (LMMSE) receiver, we derive the closed-form MSE expressions corresponding to the PU and the SU. With the derived MSEs, we propose two robust design criteria. One criterion is to minimize the ST's power consumption under the constraint that the PU's and SU's quality-of-service (QoS; i.e., MSE) can be met. The other criterion is to minimize the SU's MSE when the PU's QoS can be controlled under a certain value and the ST satisfies the limitation of its transmission power consumption. Both the optimization problems of the proposed design criteria are not convex, and the corresponding solutions cannot be directly obtained. We then propose transfering the original optimization problems into two subproblems, where each of them is eventually formulated as a convex optimization problem, and the solutions are iteratively obtained, which is effective. Thus, the results can be obtained with the interior-point method. Simulations certify the robustness of our designs. [ABSTRACT FROM PUBLISHER]