PROJECT TITLE :

A Scalable Framework for Wireless Distributed Computing - 2017

ABSTRACT:

We have a tendency to contemplate a wireless distributed computing system, in which multiple mobile users, connected wirelessly through an access point, collaborate to perform a computation task. In explicit, users communicate with every alternative via the access point to exchange their regionally computed intermediate computation results, that is called knowledge shuffling. We have a tendency to propose a scalable framework for this system, in that the required communication bandwidth for information shuffling will not increase with the number of users in the network. The key idea is to utilize a explicit repetitive pattern of placing the data set (so a specific repetitive pattern of intermediate computations), so as to provide the coding opportunities at each the users and the access purpose, which scale back the required uplink communication bandwidth from users to the access purpose and the downlink communication bandwidth from access purpose to users by factors that grow linearly with the amount of users. We have a tendency to also demonstrate that the proposed knowledge set placement and coded shuffling schemes are optimal (i.e., achieve the minimum needed shuffling load) for both a centralized setting and a decentralized setting, by developing tight info-theoretic lower bounds.