Graduate Thesis Or Dissertation
 

Towards Long Short Term Memory Based Proactive In-Network Caching for Cloud RANs

Public Deposited

Downloadable Content

Download PDF
https://ir.library.oregonstate.edu/concern/graduate_thesis_or_dissertations/mk61rp85s

Descriptions

Attribute NameValues
Creator
Abstract
  • The rapid population growth in large urban cities has led to an unprecedented increase in both the number and the diversity of wireless devices and applications with varying quality of service requirements in terms of latency and data rates. LinkNYC is an example of an urban communication network infrastructure, which replaces all the payphones in the five boroughs of New York City (NYC) with kiosk-like structures, called Links, with the goal of bringing fast and free public Wi-Fi access to thousands of city users. When enabled with data storage capability, these Links can play the role of edge cloud devices to allow in-network caching of popular Internet content to reduce access delay and backhaul traffic congestion by placing content closer to the end-users. In this thesis, we propose k-means clustering to optimize content placement at the BSs, and we do so by clustering BSs based on the transmission delay between BSs and by proactively caching content at the clustered BSs with the aim of reducing content access delay and traffic congestion. Our proposed scheme also uses the Long Short Term Memory (LSTM)'s Seq-to-Seq model to predict content popularity at each BS. Using the LinkNYC network topology as the use case, we show through simulations that the proposed scheme reduces content access delay by minimizing content miss ratios and by reducing in-network content redundancy. Our study shows that our hybrid approach of proactive and reactive caching coupled with LSTM based popularity prediction provides potential solutions for fulfilling growing demands in urban communication networks.
License
Resource Type
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Committee Member
Academic Affiliation
Rights Statement
Publisher
Peer Reviewed
Language
Embargo reason
  • Intellectual Property (patent, etc.)
Embargo date range
  • 2019-11-16 to 2020-12-17

Relationships

Parents:

This work has no parents.

In Collection:

Items