Smart execution strategy selection for multi tier execution in Named Function Networking

Abstract

Named Function Networking was designed to perform computations inside Information Centric Networks, whereas the network decides where to execute a computation depending on the input data. However, this decision is based on fixed forwarding strategies. Initially, the forwarding strategies aimed to execute computations deep inside the network satisfying the requirements of data centers and cloud computing facing large input data and computational intensive operations. Data are produced by content providers, later processed and delivered from the cloud towards the clients. Recently, this characteristic changed as Internet of Things (IoT) devices entered the game. Sensors for home automation, autonomous cars, etc generate new data, at the edge of the network. To respond to those developments, forwarding strategy can be modified to perform computations at the edge of the network to provide low latency and to reduce the traffic to the core of the network. However, this reduces the capabilities of the network to choose the execution location and a computation may require input data from the edge as well as from data centers. For those cases, neither computing at the edge nor in the cloud is sufficient. By enabling NFN nodes to decompose computations and to choose a forwarding strategy per subcomputation, edge computing nodes can offload parts of the computation into the network for execution inside data centers.