论文标题
通过多播路由网络编码的缓存
Coded Caching over Multicast Routing Networks
论文作者
论文摘要
Maddah-Ali和Niesen（MAN）最初提出的编码缓存方案将编码的多播消息从服务器传输到通过电容的共享链接配备缓存的用户，并被证明是在恒定多重因子中的理论上最佳信息。这项工作将MAN方案扩展到一类两跳有线无线网络，包括通过Fronthaul链接连接到$ H $ HUB Helper节点（接入点/基站）的一层的一台服务器，该层依次通过无线访问网络与$ K $用户进行通信，每个都配备了自己的Cache。考虑了两个变体，在访问段的建模上有所不同。对于物理场景，例如局部网络和蜂窝网络，应将两种模型视为网络层的抽象，在某个覆盖区域上分布在空间上。我们方法的关键重点是通过网络路由Man-type多播消息，并将最佳路由方案制定为可以精确解决的优化问题，或为我们提供强大的启发式算法。 Our approach solves at once many of the open practical problems identified as stumbling blocks for the application of coded caching in practical scenarios, namely: asynchronous streaming sessions, finite file size, scalability of the scheme to large and spatially distributed networks, user mobility and random activity (users joining and leaving the system at arbitrary times), decentralized prefetching of the cache contents, end-to-end encryption of HTTPS请求，它使助手节点忽略了用户的要求。
The coded caching scheme originally proposed by Maddah-Ali and Niesen (MAN) transmits coded multicast messages from a server to users equipped with caches via a capacitated shared-link and was shown to be information theoretically optimal within a constant multiplicative factor. This work extends the MAN scheme to a class of two-hop wired-wireless networks including one server connected via fronthaul links to a layer of $H$ helper nodes (access points/base stations), which in turns communicate via a wireless access network to $K$ users, each equipped with its own cache. Two variants are considered, which differ in the modeling of the access segment. Both models should be regarded as abstractions at the network layer for physical scenarios such as local area networks and cellular networks, spatially distributed over a certain coverage area. The key focus of our approach consists of routing MAN-type multicast messages through the network and formulating the optimal routing scheme as an optimization problem that can be solved exactly or for which we give powerful heuristic algorithms. Our approach solves at once many of the open practical problems identified as stumbling blocks for the application of coded caching in practical scenarios, namely: asynchronous streaming sessions, finite file size, scalability of the scheme to large and spatially distributed networks, user mobility and random activity (users joining and leaving the system at arbitrary times), decentralized prefetching of the cache contents, end-to-end encryption of HTTPS requests, which renders the helper nodes oblivious of the user demands.
