Information Resilience in a Network of Caches With Perturbations

Caching in a network of caches has been widely investigated for improving information/content delivery efficiency (e.g., for reducing content delivery latency, server load and bandwidth utilization). In this work, we look into another dimension of network of caches - enhancing resilience in informat...

Full description

Saved in:
Bibliographic Details
Published in:IEEE access 2021, Vol.9, p.136135-136150
Main Authors: Wu, Dehao, Chai, Wei Koong
Format: Article
Language:English
Subjects:
Algorithms
Approximation algorithms
Availability
Caching
Collaboration
Failure rates
Information dissemination
Information resilience
Maximization
Network latency
network of caches
Network topologies
Network topology
network with perturbations
Nodes
Optimization
Perturbation
Perturbation methods
Resilience
Servers
Topology
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Caching in a network of caches has been widely investigated for improving information/content delivery efficiency (e.g., for reducing content delivery latency, server load and bandwidth utilization). In this work, we look into another dimension of network of caches - enhancing resilience in information dissemination rather than improving delivery efficiency. The underlying premise is that when information is cached at more locations, its availability is increased and thus, in turn, improve information delivery resiliency. This is especially important for networks with perturbations (e.g., node failures). Considering a general network of caches, we present a collaborative caching framework for maximizing the availability of the information. Specifically, we formulate an optimization problem for maximizing the joint utility of caching nodes in serving content requests in perturbed networks. We first solve the centralized version of the problem and then propose a distributed caching algorithm that approximates the centralized solution. We compare our proposal against different caching schemes under a range of parameters, using both real-world and synthetic network topologies. The results show that our algorithm can significantly improve the joint utility of caching nodes. With our distributed caching algorithm, the achieved caching utility is up to five times higher than greedy caching scheme. Furthermore, our scheme is found to be robust against increasing node failure rate, even for networks with a high number of vulnerable nodes.
ISSN:2169-3536
2169-3536
DOI:10.1109/ACCESS.2021.3111699