In-network packet-level caching for error recovery in ICN

In-network packet-level caching is one of the salient features of Information-Centric Networking (ICN) architectures, offering reduced communication latency, sender load, and network traffic. The literature on ICN caching is quite extensive, exploring in depth different aspects of caching, such as the positioning of storage resources in the network, the caching policies of individual caches, as well as the interaction of caching and routing. However, many researchers have questioned the value of in-network caching, arguing that the limited storage resources of ICN routers cannot eliminate a significant amount of redundant transmissions, hence the performance advantages of in-network caching may be similar to those of traditional edge-caches. Nevertheless, there is another potential use of in-network caching that poses less challenging storage requirements: exploiting it to enhance the efficiency of error recovery by speeding-up retransmissions of lost packets. In this paper, we investigate the impact of in-network packet-level caching on transport-layer error recovery. We first sketch a retransmission-oriented packet-level caching scheme, highlighting the differences in the performance requirements compared to general-purpose caches. Then we model and assess its expected gains in terms of latency and network traffic reduction, and finally we provide a reality check of the approach based on current network requirements and hardware specifications.