Savant: An Accounting and Accountability Framework for Information Centric Networks
Citation:
Diarmuid Collins, Savant: An Accounting and Accountability Framework for Information Centric Networks, Trinity College Dublin, 2016Download Item:
dcollins-Final-corrected-PhD-Thesis.pdf (PDF) 4.652Mb
Abstract:
Content Provider i.e. entities that own or are licensed to sell and distribute content e.g., HBO,
Netflix, Amazon Prime, which use the IP-based Internet model for content distribution, consume a large percentage of network bandwidth that is expected to almost treble between 2014 and 2019 [Cisco, 2015]. This requires significant infrastructural investment by content providers, content distributors (i.e. fixed-infrastructure content distribution networks (CDNs)) and network operators e.g., AT&T. Consequently, content providers are looking for more efficient, cheaper, secure, scalable and accountable mechanisms for the distribution of content to end-users. Many existing and proposed future content distribution architectures offer desirable elements that may lead to less bandwidth usage, reduced network congestion, higher content availability and reduced costs e.g., multicast IP, peer-to-peer (P2P), and so forth. For example, the Information Centric Networking (ICN) paradigm offers solutions to many of these challenges by decoupling user trust in content from where it is obtained by enabling the content to self-verify i.e. the user can establish integrity, trust and provenance in content received from trusted or untrusted infrastructure. However, these architectures typically span domains of trust that lack central administration. Consequently, it is difficult to gather reliable accounting and accountability information for the content distribution process, which we argue is a fundamental business requirement for many content providers. Content accounting refers to any information that a content distributor needs to track, relating to the delivery of content to its intended consumers. In contrast, content accountability refers to the willingness of the communicating infrastructure to produce accurate and verifiable information about the content distribution process. The primary difference between an accounting architecture and accountability architecture is that when trust fails the latter has the tools to pinpoint the responsible entity with non-repudiable evidence. In this thesis, we develop two tools to help identify the drawbacks and merits of existing architectures. The first is a taxonomy for accounting information based on our analysis of logging information gathered from the surveyed systems. The second is a generic model for content distribution based on a synthesis of desirable elements from the surveyed architectures. Utilising these tools, we propose an ICN architecture extension for content accounting and accountability called the Savant framework, which we apply to the Named Data Networking (NDN) architecture. Savant naturally supports efficient content distribution while gathering non-repudiable near real-time information efficiently from NDN clients and NDN caches. This is supported using NDNs natural support for security, integrity and trust and by maintaining hash chains of logs and commitment to log integrity between communicating nodes. Our proof-of-concept implementation, which is based on an NDN video content distribution session, demonstrated that accounting and accountability information can be gathered for an ICN packet level architecture. Our analysis also showed that the overhead on the system is very small if several extensions and improvements are applied to the architecture for efficiency. As described, ICNs/NDN with Savant support could eventually complement or replace today’s CDN infrastructure with a scalable, trustworthy, reliable accounting and accountability framework for content distributed in trusted and untrusted environments meeting the diverse requirements of content providers, network operators and end-users.
Sponsor
Grant Number
TGI-3 program
HEA-PRTLI Cycle 3 and Cycle 4 (CSCB)
Author's Homepage:
http://people.tcd.ie/dcollin5Description:
PUBLISHED
Author: Collins, Diarmuid
Advisor:
O'Mahony, DonalQualification name:
Doctor of Philosophy (Ph.D.)Publisher:
Trinity College (Dublin, Ireland). School of Computer Science & StatisticsType of material:
ThesisCollections:
Availability:
Full text availableSubject (TCD):
Telecommunications , Accounting and accountability Information centric networking , Pervasive ComputingDOI:
10.13140/RG.2.2.18326.96326Licences: