QoS of mesh networking: Difference between revisions

From OLPC
Jump to navigation Jump to search
No edit summary
m (typos)
Line 1: Line 1:
==== Current situation ====
==== Current situation ====


Nowadays trafic controlling in mesh networking is not quite solved. Network can act well when there are not many laptops. But the more there are machines the more difficult it is to control trafic and the mere possibility of message passing.
Nowadays traffic controlling in mesh networking is not quite solved. Network can act well when there are not many laptops. But the more there are machines the more difficult it is to control trafic and the mere possibility of message passing.


Usual way of trafic controlling is observing part of network without taking into account the whole one. Unfortunatelly when a network is wide there can be problems. First of all it can lead to a non-optimal path selection and routing. That can cause congestions and not shortest paths.
Usual way of trafic controlling is observing part of network without taking into account the whole one. Unfortunately when a network is wide there can be problems. First of all it can lead to a non-optimal path selection and routing. That can cause congestion and not-shortest paths.


Another varient is to control the whole network. This means not only developing more sophisticated algorithms but also thinking about the time of processing. Such algorithms should still be quick enough to provide QoS of high level. Another side of the problem are limited laptop resources.
Another varient is to control the whole network. This means not only developing more sophisticated algorithms but also thinking about the time of processing. Such algorithms should still be quick enough to provide QoS of high level. Another side of the problem are limited laptop resources.
Line 25: Line 25:
==== Perspectives ====
==== Perspectives ====


This project can be used later to arrange mesh networking with common computers what can cause larger trafic than in XO laptop network.
This project can be used later to arrange mesh networking with common computers what can cause larger traffic than in XO laptop network.

Revision as of 21:00, 31 March 2008

Current situation

Nowadays traffic controlling in mesh networking is not quite solved. Network can act well when there are not many laptops. But the more there are machines the more difficult it is to control trafic and the mere possibility of message passing.

Usual way of trafic controlling is observing part of network without taking into account the whole one. Unfortunately when a network is wide there can be problems. First of all it can lead to a non-optimal path selection and routing. That can cause congestion and not-shortest paths.

Another varient is to control the whole network. This means not only developing more sophisticated algorithms but also thinking about the time of processing. Such algorithms should still be quick enough to provide QoS of high level. Another side of the problem are limited laptop resources.

Ideas

In modern world the requirements only grow. Now we need the posibility to arrange mesh networks with large number of nodes (laptops). The main idea is to arrange such networks with regard to Quality of Service. This means both providing reserving the resources and guaranteed service quality.

Main goals

My aim is to create the technique of effective mesh networking for XO laptops. QoS is the main goal.

Main features of the technique to provide QoS are:

  • taking into account the whole network architecture
  • finding the shortest path
  • congestion control
  • high speed of networking algorithms
  • providing high level of quality of service
  • resources distribution and other necessary for QoS things

Perspectives

This project can be used later to arrange mesh networking with common computers what can cause larger traffic than in XO laptop network.