17-03-2014, 09:34 PM
Abstract: Qos Anycast is a newly rising network
communication service, playing an important role in the area
of video stream on-demand. Although the DiffServ network
QoS Anycast Routing architecture is very good, the
interleaving services problems exist, which makes its service
efficiency decline. A QoS Anycast network cluster balancing
algorithm based on genetic algorithm and scheduler is
introduced in this paper. A real-time feedback information on
scheduler and server is designed with the program of
dynamically adjusting server weights on genetic algorithms
program, to accomplish the anycast network cluster balancing
choice based on real-time feedback for better utilization
efficiency of the server and throughput of cluster system.
Keywords: QOS, video teaching, genetic algorithm, network
cluster
Compared to the traditional narrow-band communication,
video application has a higher demand on the
communication network, especially when the data networks
adopt IP technology more and more. It is critical about how
to meet the QoS requirements of video applications.
Anycast is a model of network communications services.
IPv6 defines it with unicast and multicast, which is one of
the three kinds of network services in the future [1]. Anycast
can put multiple video streaming servers on the different
places of the network to reduce the occupation of network
resources, increase service scalability, and meet multi-user
QoS. In the research of Anycast Routing architecture, even
DiffServ network QoS Anycast Routing architecture is very
good, the interleaving services problems exist, which makes
its service efficiency decline.
A QoS Anycast network cluster balancing algorithm
based on genetic algorithm and scheduler is introduced in
this paper. A real-time feedback information on scheduler
and server is designed with the program of dynamically
adjusting server weights on genetic algorithms program, to
accomplish the anycast network cluster balancing choice
based on real-time feedback for better utilization efficiency
of the server and throughput of cluster system.
1 Qos Anycast network cluster balancing design based on
Genetic Algorithm
GA (Genetic Algorithm) is a newly developed global
optimization algorithm in recent years, which achieves the
improvement of individual adaptation for the optimization
of the resources allocation in the view of biological genetics
through the mechanisms like natural selection, inheritance,
and variation. Genetic algorithm is an adaptive global
optimization probability search algorithm through
simulating the inheritance and evolution of biology in
nature [2], which is Darwin's calculation model for
biological evolution of genetic and natural selection. It
stems from biology genetics and law of the survival of the
fittest, which is the search algorithm with "survival +
detection" iterative process.
Genetic algorithm regards every individual in a group as
object, and search efficiently in an encoding parameter
space with the random technical guidance [3]. The selection,
crossover and mutation constitute the genetic manipulation
of the genetic algorithm; coding parameters, the initial
group settings, the fitness function design, genetic
manipulation design, and control parameters settings
comprise the core of the genetic algorithm. The basic
process is shown as below: