17
MATHEMATICAL EQUATIONS THAT CAN IMPROVE NETWORK SPEEDS 1 K S KIRAN CS13B1018

N A Riksk

  • Upload
    nikhith

  • View
    215

  • Download
    0

Embed Size (px)

DESCRIPTION

Same as above.

Citation preview

  • MATHEMATICAL EQUATIONS THAT CAN IMPROVE NETWORK SPEEDS

    1

    K S KIRANCS13B1018

  • KEY WORDS RELATED Network coding(random linear)

    KODO

    Mathematical equations

    Meshed networks

    On the fly approach2

  • LIST OF READINGSResearch papers on:

    Dynamic adaptive streaming: A dynamic adaptive protocol FRANC is tested on various TCP versions and proved to be faster than existing protocol HTTP DASH.

    Sub-transport layer coding: Uses a coding shin that aids IP layer traffic in lossy environments while being transparent to transport layer protocols.

    3

  • LIST OF READINGS Reed Solomon Protocol: Based on univariate

    polynomials over finite fields, it is an error correction code.

    4

  • ABOUT

    A stateless transmission approach that avoids retransmission and error correction using on the fly encoding and decoding.

    Using codes containing mathematical equations of packets to achieve higher network speeds.

    5

  • ABOUT

    The technique processes data packets in chunks thus allowing users to receive them in any order.

    It is especially useful in multi-media content transmission and establishing secure connections.

    7

  • GOAL

    Linear network coding is optimised except for Gauss-Jordan elimination and linear dependency among coefficient vectors.

    Working on a better coding of packets.

    9

  • MAKING PROGRESS

    Getting familiarised with KODO, a RLNC library.

    Testing existing algorithms and working on new ones.

    Getting familiarised with mini net.

    10

  • FURTHER READINGSSDN using open flow (PPT circulate in IITH).

    Computer networking a topdown approach by Keith & Ross.

    SSL and its Commands.

    Different RLNC Methods and their advantages in different situations.

    Eg: Seed based, Perpetual, Fulcrum11

  • IMPLEMENTATION DETAILS

    Worked on some sample codes on RLNC.

    Specific to encoding : A randomly generated matrix is multiplied to packet matrix.

    decoding : Inversion is made through Gauss - Jordan elimination and

    re-encoding.

    12

  • INTERESTING APPLICATIONSUsing multicast to achieve synchronisation.

    Local sharing of pictures from one sender to many receivers.

    13

  • 14

  • PAPER EXCLUSIVE

    Just have been reading a few similar papers to get the idea of viewing an existing concept in a different perspective.

    One such is improvement of Gauss Jordan elimination used in coding/decoding of packets.

    15

  • MISCELLANEOUS

    For the sake of implementation a concept called parametrized inheritance(Mixin - Layers) in C++ is necessary and reading through it.

    Started reading the documentation of KODO.

    16

  • THANK YOU

    17