InfoMax: A transport-layer paradigm for the age of data overload

Traditional transport-layer protocols, such as TCP and UDP, envisioned a communication architecture with receivers who need all (or at least most) of the data originating at the respective senders. In the emerging age of data overload, a more common application data-access pattern will increasingly be that of consumers who need a representative sampling of data at producers, leading to a paradigm shift in transport protocols. InfoMax implements this abstraction on top of information-centric networks (ICNs). In this chapter, we present the design of InfoMax, a new transport-layer protocol that allows applications to request a representative sampling of a named data set. InfoMax exploits the recently proposed information-centric network paradigm that gives hierarchical names to data items, as opposed to IP addresses. Assuming that named objects that share a longer branch in the name space are semantically more similar, InfoMax has the property of minimizing semantic redundancy among the data items selected for delivery, hence offering the best coverage of the requested data set with the fewest delivered items. This chapter discusses the design of InfoMax, its experimental evaluation, and example applications.