Dynamic computations in ever-changing networks
Idit Keidar · 2011
This talk focuses on dynamic computations (sometimes called live, on-going, continuous, or stabilizing), which continuously adapt their output to reflect input and network topology changes. Three specific examples are discussed: continuous weighted matching, live monitoring, and peer sampling (also called gossip-based membership). Such computations are of interest in ever-changing networks, where the network topology itself (nodes and links) constantly changes, as do the inputs to the computation, e.g., sensor reads. Everchanging networks occur in many settings nowadays, including ad-hoc, vehicular, and sensor networks, social networks, and clouds spanning multiple data-centers.