Citation
Huyop @ Ayop, Fahrul Hakim
(2016)
Traffic management algorithms for LEO satellite networks.
Doctoral thesis, Universiti Putra Malaysia.
Abstract
Traffic management for Low Earth Orbit (LEO) satellite networks deals with the
process of monitoring the network activities by allocating certain traffic into the right
path and increasing the throughput rate for the sake of performance and efficiency.
This thesis deals with traffic management by improving some algorithms in routing
and congestion avoidance to guarantee the subscribers to have their desired QoS.
The problem in network arises during the period of coexistence between UDP and
FTP traffic in the same network which could affect their performances. The situation
is getting worse when the traffic is not equally distributed across the network which
could possibly degrade the network performance especially in the case of the
delivery of the high priority traffic. A failure to manage traffic classes by routing
them according to their type of service could also bring negative impact to LEO
satellite network performance.
In dealing with traffic routing problem, two algorithms, Dijkstra's Shortest Path and
Genetic Algorithm (GA) are combined together and enhanced to re-strategize a
better routing mechanism for a heterogeneous mix of traffic classes ranging from
traditional voice calls to multimedia data services in Low Earth Orbit (LEO) satellite
networks. Those two algorithms are combined together due to the strong possibilities
of being unable to achieve the optimal results when implemented separately. Three
classes of traffic such as low, medium and high priority are defined and to be
allocated to the right path from source to destination with the most privileged is given
to the high priority traffic.
In satellite network with Multi Protocol Label Switching (MPLS), routing with
Extended Dijkstra Shortest Path algorithm is done to improve QoS by minimizing
the link shared between UDP and TCP traffic flows. By minimizing the link shared
between those traffics, the performance of UDP traffic which carries delay sensitive
data could be improved.
Since TCP WestwoodNew is designed to be implemented in wired and wireless
network environment, there are few drawbacks when TCP WestwoodNew is
implemented in the satellite network environment. As an example, the sender cannot
fully utilize the available bandwidth because the rate of the congestion window
increment in Slow Start phase of TCP WestwoodNew is rather slow. In this thesis,
congestion avoidance algorithm of TCP WestwoodNew is enhanced in order to
improve the drawbacks.
In this thesis, several proposed algorithms have been developed to improve the traffic
performance in the LEO satellite network. In order to evaluate the proposed
algorithm, a series of experiments to implement Discrete Event Simulation (DES) of
a LEO satellite network by using ns-2 and C/C++ are conducted. The performances
of the proposed algorithms are then compared with the previously developed
algorithms. The important parameters measured in the simulation are delay rate,
throughput rate and fair traffic distribution rate. From the results, the proposed
algorithms have managed to reduce the average delay rate for the high priority and
UDP traffic. The results also indicate that there is an improvement of fair traffic
distribution rate in the network where most of the Inter Satellite Links (ISLs) are
able to maintain the link loading percentage ranging from 25% to 75%. For the TCP
traffic performance, there is 5% improvement in terms of throughput rate increment.
Overall, from all the three experiments that have been conducted, the proposed
routing algorithms which have been developed to manage the network traffic have
proved better than the previously developed algorithms in several perspectives
especially for the implementation of real-time applications in LEO satellite
networks.
Download File
Additional Metadata
Actions (login required)
|
View Item |