Reconfigurable intelligent surface (RIS) has emerged as a prospective technology, capable of shaping radio wave propagation and enhancing performance gains. The RIS aided multiple input multiple output interference channel is considered. We aim to maximize the sum rate by jointly optimizing the precoding matrices and RIS parameters, subject to the transmit power constraints. Both upper and lower bound sum rate approximation schemes are designed for this nonconvex problem. First, for the upper bound approximation, the minimax problem is formulated through the approximated Lagrangian function, where the precoding matrices are determined by the RIS matrix and the Lagrange multiplier, and are eliminated in the problem. A single loop primal dual algorithm is proposed. In each iteration, the RIS parameter and the Lagrange multiplier are updated by one projected gradient step and quadratic interpolation, respectively. Its complexity only grows linearly in the number of RIS elements. In addition, the total signal to total interference plus noise ratio is introduced for the sum rate lower bound approximation. The fractional objective function is reformulated via the Dinkelbach’s technique, and the variables are updated with closed form through alternating optimization and projected gradient method. The proposed two approximation schemes and methods are also extended to the general multi-RIS multi-cell network with multiple users. Simulations show that the upper bound approach performs well with only 10% computational time of the compared methods, and shows high efficiency in one-iteration test; the lower bound approach achieves almost the highest sum rate using little computational cost.

The multi-user multiple-input multiple-output interference channel assisted by reconfigurable intelligent surface (RIS) is considered. We jointly optimize the precoding and decoding matrices as well as RIS discrete parameters, to maximize the desired signal power with interference alignment (IA) conditions. The interference leakage is penalized to the objective function by the Courant penalty technique. This simplifies the optimization problem and tackles the possible infeasibility. In the new objective function, the desired signal power maximization and the interference leakage minimization are combined together. We propose two efficient strategies to update the penalty parameter, namely the constraint violation update and the total signal to total interference plus noise ratio (TSTINR) update. The variables are updated alternatively, with either Schur decompositions or projected gradient inner updates. Simulations show that the proposed model greatly improves the benchmark model in terms of the achieved sum rate with less computational cost.

With the increasing communication and storage of point cloud data, there is an urgent need for an effective objective method to measure the quality before and after processing. To address this difficulty, we propose a projection-based blind quality indicator via multimodal learning for point cloud data, which can perceive both geometric distortion and texture distortion by using four homogeneous modalities (i.e., texture, normal, depth and roughness). To fully exploit the multimodal information, we further develop a deformable convolutionbased alignment module and a graph-based feature fusion module, and investigate a graph node attention-based evaluation method to forecast the quality score. Extensive experimental results on three benchmark databases show that our method achieves more accurate evaluation performance in comparison with 12 competitive methods.