Neighbour-based on-demand routing algorithms for mobile ad hoc networks
Mobile Ad hoc NETwork (MANET) is a set of wireless mobile nodes temporary connected without existing network infrastructure or centralized administrations. These nodes with random movement and limited resources have created quite a number of new challenging research issues. One such issue is the...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2017
|
| Subjects: | |
| Online Access: | http://psasir.upm.edu.my/id/eprint/68750/1/FSKTM%202018%2010%20IR.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Mobile Ad hoc NETwork (MANET) is a set of wireless mobile nodes temporary
connected without existing network infrastructure or centralized administrations.
These nodes with random movement and limited resources have created
quite a number of new challenging research issues. One such issue is the routing,
which has recently received significant attention from many researchers.
In particular, the problems of the dropping decision, routing overhead and
network density in the route request stage.
The routing overhead is due to the broadcasting method used in route discovery,
which floods the network with a Route REQust message (RREQ). Accordingly,
the aim of this research is to address the deficiency of the broadcasting method
at route request stage.
To overcome the issues related to the broadcasting method, several routing
algorithms have been proposed over the Ad hoc on demand Distance Vactor
(AODV) such as Neighbor Coverage-Based Probabilistic Rebroadcast (NCPR).
Although the NCPR did overcome the AODV routing algorithm in terms of
reducing the routing overhead, such algorithm has its drawbacks. Thus, there is
a room for further enhancement to develop a routing algorithm as to mitigate
the drawbacks with the NCPR.
The dropping decision of the redundant RREQ in the NCPR algorithm completely
relies on preset variables, such variables require to be set by the system
administrator based on the scenario. Unfortunately, the setting which is proper
for a specific scenario is not suitable for another. Furthermore, the connectivity factor which is used to estimate the connectivity ratio at each node is still unable
to estimate such ratio accurately. Therefore, a Dynamic Connectivity Factor
Probabilistic (DCFP) is proposed, based on a novel formula that dynamically
adjusts the dropping decision based on the neighbor information gathered from
the node itself.
As the number of nodes or the network traffic load increases, the NCPR fails to
relieve the routing overhead due to the increase in the RREQ redundant messages.
Thus, a Scalable Neighbor-Based Routing algorithm (SNBR) is proposed,
which reduces the routing overhead in the NCPR, by eliminating the redundant
RREQ. The broadcasting in this algorithm is governed by the inverse relation
between the number of neighbors and the probability of the rebroadcasted
RREQ messages. This algorithm enhances the network performance, even
though the network is experiencing an increase in the number of nodes or traffic
load.
Naturally, nodes in MANET are free to move forming an arbitrary topology
and at any time the network density may change . Such change may lead to
an extreme performance degradation, especially when the routing algorithm
relies on fixed threshold values in the dropping decision. Accordingly, a
Novel Density-Aware Routing algorithm (NDAR) is proposed. The proposed
algorithm totally eliminates the need for fixed threshold values through the use
of a novel formula that can easily estimate the node density and replace the
fixed threshold value based on the neighbor information.
All the three proposed algorithms are evaluated using discrete event simulation,
in particular Network Simulator tool (NS2), and compared with the latest
routing algorithm (NCPR ) and fundamental algorithm (AODV) using five performance
metrics. The first algorithm DCFP outperforms the NCPR algorithm
in terms of normalize routing overhead by 11.27%, while maintaining the same
packet delivery ratio. In addition, with regard to the second algorithm SNBR, the
results show that SNBR overcomes the NCPR algorithm terms of normalize routing
overhead by 58.80% as its due to its dropping factor. Furthermore, the third
algorithm NDAR presents further enhancement and better performance in all
five performance metrics as compared to NCPR and AODV algorithms in a low
or high density of nodes. In terms of the applications, The DCFP is more suitable
to be used for education applications, while the SNBR is a good algorithm
designed to be used for rescue system as data and energy is the main concern.
Finally, the NDAR is more suitable for personal area and home networking. |
|---|