prospective use of bloom filters and muxing for information centric network caching

Post on 15-Apr-2017

316 Views

Category:

Science

2 Downloads

Preview:

Click to see full reader

TRANSCRIPT

PROSPECTIVE USE OF BLOOM FILTER AND MUXING FOR INFORMATION CENTRIC NETWORK CACHING

Ibrahim Abdullahi, Suhaidi Hassan, Suki Arif

2 0 1 4

PRESENTATION OUTLINE

Introduction & Motivation1

2 Information Centric Network Overview &Challenges and Research issues

3 Bloom Filters and Muxing

Proposed Concept4

INTRODUCTION

Information Centric Network (ICN) has seen different folds in recent years.

GOAL: •Communication and information referral on the Internet by names in place of IP addresses.

•A neighbor node provides information once cached.

•Fast way of information sharing.

•Lesser bandwidths, time and less delays (Latency)

MOTIVATION

Information Centric Network (ICN) like other paradigms has yearn to mitigate some Internet challenges.

•Before the coming of IPV6, addresses will soon be exhausted.

•High demands of Multimedia information on the Internet.

•Bottle neck and server overloads.

•High traffic reaching 4.8 zettabyte according to Cisco index report in 2014

Contribution

•In cooperating Bloom filters could curtails some of the challenges

ICN OVERVIEW

Information Centric Network (ICN) structure

•ICN comprises mainly of three parts: • Content Store (CS)• Pending Interest Table (PIT)• Forwarding Information Base (FIB)

•CS holds the information as a storage area. When Interests are demanded for, CS is looked up for serving the interest.

•PIT is a special interest count keeping table, it works using referred interests and also future interests. Once requests are subscribed, PIT records the interest before it is served.

•FIB handles the routing like event of interest/information forwarding to the desired node.

ICN OVERVIEW

Figure 1: ICN Overview

CHALLENGES AND RESEARCH ISSUES

Naming & Name Resolution1

2 Mobility

3 Security

4

Scheduling5

Scalability6

Caching

CACHING IN ICN

Caching: In-path Off-path

•Cache at all NODES (Redundancy)

•Coordinating Cache in-network (Leave a memory for Path movement)

•Intra-Autonomous cooperative caching (Eliminating redundancy but incurred high searching)

*

BLOOM FILTERS AND MUXING

Bloom Filters: A data structure algorithm used to

minimize space and data placement

Properties of Bloom Filters

•Fast look-ups during search

•Returns false positive

•Easily identify a member of a set or not

*

MUXING

Muxing is done using three distinct operations

1. Data is sent to the network by the User

2. Data is compared by an Agent

3. Data is then MuXed

… BLOOM FILTERS AND MUXING

*

Muxing is possible when the interests from several Subscribers are AND (intersected).

Figure 2: Muxing Interest

PROPOSED CONCEPT

Figure 3: Proposed Concept

…SCENARIO

Scenario descriptionSuppose a populated network is set as Z:1: Subscriber S request an interest Yi, such that i is the unique identity of the interest2: i is placed on the PIT of the first node3: Look-up from the first Router node R1 to get a match using bloom filters4: serve the interest i from the Subscriber Yi

5: else move interest to the next router node6: Check to find the similar interests by intersecting 7: Combine the interest (mux) as one to save space and bandwidths 8: Check from the neighbor AS, the results in bloom filters9: Look into the entire network for the interest10: Publisher Xi sends an interest i , into the network11: Cache i, for the subsequent similar request on the first router node (Ri)12: Send resulting interest/object via the same route13: Cache results for next request

ADVANTAGES OF VENDING IN BLOOM AND MUXING

Overall network latency is improved

Delay jitter is lower than non-bloom filter and mux ICN architectures

Objects and interest finding is much faster

High inter-node communications to and different stations are minimized

Object redundancy is mitigated

Improved cache management through selected node caching

Easily scalable

Enhanced mobility due to the bloom element membership

Better security

Easy name resolution through bloom filter probability functions

Higher throughput by AS specification

CONCLUSION

ICN envision better naming in lieu of traditional Internet

Bloom filters mitigate the problem of large network mobility and scalability

Muxing saves higher requests from various stations

Advantages of name resolution can be achieved using Bloom

Bandwidth is saved

Q and A

top related