Edge Computing and Data Extraction

Date

2019-11-19

Journal Title

Journal ISSN

Volume Title

Publisher

item.page.doi

Abstract

The Combinatorial Optimization field consists of selecting best or optimal object or optimal set of objects from finite set of possibilities. Many computer science problems can be formulated as combinatorial optimization problems. This dissertation mainly focused on two of such problems: finding optimal offloading scheme for mobile edge computing and multi-document summarization. Due to high processing power and internet connectivity, smartphones are used widely for many applications like email, face recognition, natural language processing, interactive games etc. People prefer mobile devices to use for these applications due to their ease of use. However, because of hardware limitations, mobile devices have limited battery life, power and capacity. Researchers are constantly looking for ways to maximize the usage of these resources. To reduce the load of applications on mobile devices and use the resources efficiently, it is necessary to move some load of applications to remote servers in such a way that the applications will run seamlessly. Computation offloading for mobile-edge computing (MEC) is a mechanism to utilize resources well by moving resource-intensive units (functions, components, etc.) to edge servers at network edge. Computational offloading is formulated as a graph cut problem and a solution based on spectral graph theory is proposed. It is observed that this computation offloading problem can also be transformed as n-fold integer programming by mapping the remaining computing resources to a virtual component. To ensure reliability, a reliable shadow component scheme between multilevel servers is designed. Various computer science tasks can be modelled as data subset selection problem which deal with finding representative subset from the data based on some criteria such as importance. Some examples of data subset selection in Natural Language Processing (NLP) are extractive summarization and sub-corpus selection. In both the cases, some part of original data is filtered out. In extractive summarization, important sentences from the original document set which convey the meaning of the document set are selected. In sub-corpus selection, part of a corpus needed for a particular task at hand is selected. Combinatorial optimization deals with finding best subset after comparing different subsets using objective function. In this work, some important problems such as computation offloading and multi-document summarization are formulated as non-linear combinatorial optimization algorithms. Since these problems are NP-hard, research work is done to find some intrinsic features to propose efficient solutions.

Description

Keywords

Combinatorial optimization, Edge computing, Mobile computing, Natural language processing (Computer science)

item.page.sponsorship

Rights

Citation