Multi-antenna coded caching combines a global caching gain, proportional to the total cache size in the network, with an additional spatial multiplexing gain that stems from multiple transmitting antennas. However, classic centralized coded caching schemes are not suitable for dynamic networks as they require prior knowledge of the number of users to indicate what data should be cached at each user during the placement phase. On the other hand, fully decentralized schemes provide comparable gains to their centralized counterparts only when the number of users is very large. In this paper, we propose a novel multi-antenna coded caching scheme for dynamic networks, where instead of defining individual cache contents, we associate users with a limited set of predefined caching profiles. Then, during the delivery phase, we aim at achieving a combined caching and spatial multiplexing gain, comparable to a large extent with the ideal case of fully centralized schemes. The resulting scheme imposes small subpacketization and beamforming overheads, is robust under dynamic network conditions, and incurs small finite-SNR performance loss compared with centralized schemes.
翻译:多antenna 代码化的缓存结合了全球缓存收益,与网络总缓存规模成正比,还有来自多个传输天线的额外空间多路增益。然而,典型的中央编码缓存计划不适合动态网络,因为它们需要事先了解用户数量,才能表明在安置阶段每个用户应隐藏哪些数据。另一方面,完全分散的计划只有在用户数量非常大时才能向其集中的对口单位提供可比收益。在本文中,我们提议为动态网络制定一个新的多条码缓存计划,在动态网络中,我们不界定单个缓存内容,而是将用户与一套有限的预定义的缓存剖面图联系起来。随后,在交付阶段,我们的目标是实现混合的缓存和空间多路互存收益,这与完全集中计划的理想情况大相类似。由此产生的计划在动态网络条件下规定了小型的子包装和成型管理管理器,在动态的环境下很稳健,并且与集中计划相比,造成少量的有限-SNR性性性损失。