Immersive video, such as virtual reality (VR) and multi-view videos, is growing in popularity. Its wireless streaming is an instance of general multicast, extending conventional unicast and multicast, whose effective design is still open. This paper investigates general rate splitting for general multicast. Specifically, we consider a multi-carrier single-cell wireless network where a multi-antenna base station (BS) communicates to multiple single-antenna users via general multicast. We consider linear beamforming at the BS and joint decoding at each user in the slow fading and fast fading scenarios. In the slow fading scenario, we consider the maximization of the weighted sum average rate, which is a challenging nonconvex stochastic problem with numerous variables. To reduce computational complexity, we decouple the original nonconvex stochastic problem into multiple nonconvex deterministic problems, one for each system channel state. Then, we propose an iterative algorithm for each deterministic problem to obtain a Karush-Kuhn-Tucker (KKT) point using the concave-convex procedure (CCCP). In the fast fading scenario, we consider the maximization of the weighted sum ergodic rate. This problem is more challenging than the one for the slow fading scenario, as it is not separable. First, we propose a stochastic iterative algorithm to obtain a KKT point using stochastic successive convex approximation (SSCA) and the exact penalty method. Then, we propose two low-complexity iterative algorithms to obtain feasible points with promising performance for two cases of channel distributions using approximation and CCCP. The proposed optimization framework generalizes the existing ones for rate splitting for various types of services. Finally, we numerically show substantial gains of the proposed solutions over existing schemes in both scenarios.
翻译:虚拟现实( VR) 和多视图视频等闪烁视频正在越来越受欢迎。 其无线流是一个普通多播、 扩展常规单向和多播的事例, 其有效设计仍然开放。 本文调查了通用多播的通用速率分裂。 具体地说, 我们考虑一个多驱动器单细胞无线网络, 多连接器基站( BS) 通过一般多播向多个单线网点用户传递信息。 我们考虑在 BS 上进行线性成像, 在缓慢淡化和快速淡化的情景中, 每个用户都进行联合解码。 在缓慢淡化的情景中, 我们考虑将加权平均速率平均速率最大化。 为了降低计算的复杂性, 我们将原非连接器的预感问题分解为多个非连接器问题, 每个系统版本都提议一个。 然后, 我们提议对每个确定性框架进行一个迭代位算法, 以获得Karush- Kuhn- 速率( KKT) 快速流算法( 快速流算) 使用一个更具有挑战性的变现变现的变现的变现程序。