In this paper, we propose a new caching scheme for a random wireless device-to-device (D2D) network of nodes with local caches, where each node intends to download files from a prefixed library via D2D links. Our proposed caching delivery includes two stages, employing distributed MIMO and only communicate inside its own cluster. They showed that for a single-hop cache network, with an optimized caching placement strategy and cluster size, the network throughput scales as Θ ( ) , growing

more »

... y with the size of local √ ) hierarchical cooperations respectively. The distributed MIMO is applied to the first stage between source nodes and neighbours of the destination node. The induced multiplexing gain and diversity gain increase the number of simultaneous transmissions, improving the throughput of the network. The hierarchical cooperations are applied to the second stage to facilitate the transmissions between the destination node and its neighbours. The two stages together exploit spatial degrees of freedom as well as spatial reuse. We develop an uncoded random caching placement strategy to serve this cooperative caching delivery. Analytical results show that the average aggregate throughput of the network scales almost linearly with , with a vanishing outage probability.