Performance of MDB Routing Algorithm on Grid Topology in Communication Networks

VINetworking and Communications

Oct 6, 2011 (6 years and 17 days ago)

666 views

networks where the nodes have relative fixed positions and communicate to the Internet through one or more gateways. While traditional ad-hoc routing algorithms, such as DSR and AODV, can be used in communication networks(CN), their performance is typically less than ideal. The problem is that such algorithms make assumptions that are no longer true in CN, and those assumptions can have significant performance penalties in the CN environment. This paper studies routing algorithm named MDB (Modified Depth-Breadth routing) in grid networks.

Abstract
The grid networks are a special case of communication
networks where the nodes have relative fixed positions and
communicate to the Internet through one or more gateways.
While traditional ad-hoc routing algorithms, such as DSR and
AODV, can be used in communication networks(CN), their
performance is typically less than ideal. The problem is that
such algorithms make assumptions that are no longer true in
CN, and those assumptions can have significant performance
penalties in the CN environment. This paper studies routing
algorithm named MDB (Modified Depth-Breadth routing) in
grid networks.
Keywords
CN, MDB, Grid topology; GCN.
I. Introduction
In this paper we consider the problem of routing in communications
networks(CN): we focus on routing for Point- Point datagram
networks with grid topology and best- effort service, the most
remarkable example of such net-works being the Internet. The
goal of every routing algorithm is to direct traffic from sources
to destinations optimizing at the same time several measure of
network performance as throughput (correctly delivered bits per
time unit), packet delays and resources utilization. The general
problem of determining an optimal routing algorithm can be
stated as a multi-objective optimization problem in a non-
stationary stochastic environment. Information propagation
delays, and the difficulty to model the whole network dynamics
under arbitrary traffic patterns, make the general routing
problem intrinsically distributed. Routing decisions can only
be made on the basis of local and approximate information
about the current and the future network states.
The adaptive routing algorithm we propose in this paper,
called MDB, are distributed and mobile multiagent systems
well matching the above characteristics of the general routing
problem. The design of our algorithms has been inspired by
previous works on ant colonies and, more generally, by the
notion of stigmergy [1, 2], that is, the indirect communication
taking place among individuals through local, persistent (or
slowly changing) modifications induced in their environment.
Real ants have been shown to be able to find shortest paths
using a stochastic decision policy based only on local information
represented by the pheromone trail deposited by other ants
[3].
Algorithms that take inspiration from ants’ behavior in finding
shortest paths have recently been successfully applied to
several discrete optimization problems [2, 4 - 9]. In ant colony
optimization each one of a set of concurrent artificial ants
makes use of a stochastic local search strategy to build a
solution to the combinatorial problem under consideration. The
whole set of ants collectively search for high quality solutions
by a cooperative effort mediated by indirect communication of
information on the problem structure they collect while building
solutions.
Similarly, in MDB, artificial ants (agents) collectively solve the
routing problem by a cooperative effort in which stigmergy,
mediated by the network nodes, plays a prominent role. By
using a stochastic routing policy based on local (public) and
private information ants concurrently and asynchronously
explore the network and collect useful information. While
exploring, the ants adaptively build probabilistic routing tables
and local models of the network status using indirect and non-
coordinated communication of the information they collect.In
this paper we discuss the unique aspects of grid networks, and
their differences from other networks. In particular, we propose
an algorithm for routing in such networks that is able to take
advantage of the capabilities of such networks that are not
present in ad-hoc networks. We provide some evidence that
the approach we propose is likely to perform noticeably better
than existing adhoc routing protocols.
II. Grid Topology
Grid networks have the potential to play a critical role as an
alternative technology for last-mile broadband Internet access.
They can be viewed as special case communication multi-hop
ad-hoc networks, in which each node operates both as a host
and as a router. However, GCNs (grid communication networks)
have a number of features that distinguish them from pure
ad-hoc networks. First, the positions of different nodes of a
GCN are relatively fixed. By relatively fixed position, we mean
that, although the nodes may not be absolutely immobile,
any change of position is limited within certain range. The
implication of this is that routing paths can be created that
are likely to be stable. This substantially reduces the need for
routing packet overhead. Indeed, such routing packets are
likely only needed at initialization and when traffic volume is
sufficiently low that a node cannot be sure that its neighbor is
still present, as opposed to having crashed. Second, unlike pure
ad-hoc networks, where the traffic flows between arbitrary pairs
of nodes, in GCN, all traffic is either to or from a designated
gate-way, which connects the grid network to the Internet.
The relevance of this point is that the traffic may be split over
multiple gateways, so as to reduce the load within any given
portion of the network. Third, the nodes will typically have
access to a power source, and so power consumption is not
a critical issue. Finally, such systems can be created within a
single domain of authority, and so many security issues present
in ad hoc networks are no longer relevant.
Grid communications networks are, easy to install. The setup
cost for Internet service providers (ISPs) is only gateway
installation and configuration. This makes GCN a good choice
compared to traditional directional antenna wireless access.
Scalability is a second advantage for GCNs. When new sub-
scribers activate their Internet connections, ISPs only need to
perform an authentication process to decide whether to admit
or deny. Nodes can be added one at a time, and the more nodes
admitted, the more reliable the network, because a densely
distributed network tends to maintain higher connectivity.
Traditional directional antenna networks, on the other hand,
Performance of MDB Routing Algorithm on Grid Topology
in Communication Networks
1
G.Raghu Ram,
2
S.Anuradha,
3
Dr. V.Raghunatha Reddy,
4
Dr. K.E.Sreenivasa Murthy,
5
M.Sarika
1,2,5
MCA Dept, G. Pulla Reddy Engineering College, Kurnool, Andhra Pradesh, India
3
S.K.University, Ananthapur, Andhra Pradesh, India
4
SVIT, Ananthapur, Andhra Pradesh, India

I
nternatIonal
J
ournal

of
C
omputer
S
CIenCe

and
t
eChnology
83
I S S N : 2 2 2 9 - 4 3 3 3 ( Pr i nt ) | I S S N : 0 9 7 6 - 8 4 9 1 ( Onl i ne )
www.i j c s t.c o m
IJCSt V
ol
. 2, I
SSue
1, m
arCh
2011
suffer from poor scalability since, when a new subscriber is
admitted, the antenna’s direction has to be manually adjusted.
Further-more, if a new gateway is installed to alleviate heavy
traffic, half of existing subscribers’ antennae should be reamed
to new gateway, putting a heavy cost on ISPs. The grid network
topologies were chosen as an example of a ‘large’ network in
which it is easy to investigate and describe traffic, since it has
a regular structure. We wanted a regular topology because of
the ‘reduced’ complexity and simply because it is easier to get
a general view of the paths in the network.
A. Topology with 16 nodes ( 4X4 grid )
We had furthermore performed experiments on a 4X4 grid
network, which of course consists of 16 routers connected in
a grid with wires (with a delay of 10) between adjacent routers.
This topology is used to test MDB ability to route in a ‘larger’
(but still simple) network.
Fig.1: Grid topology with 16 nodes
III. Modifications Proposed to DB Routing Algorithm
The original DB routing algorithm focuses only on the searching
method used in finding out the shorter paths in the network
but not on node balancing concept.Based on the original DB
routing algorithm few modifications are suggested to be added
to enhance its performance.
A. Congessition Control (Control of the Number of Ants
in the Network)
For every algorithm, the network load generated by routing
packets is stored as the ratio between the bandwidth occupied
by all the routing packets and the total available network
bandwidth. The routing overhead is the main function of the
topological properties of the network and of the generation rate
of the routing information packet. AntNet produces a routing
overhead depending on the ants’ generation rate and the
length of the path along they travel. As the followed path of
routing ant grows (either because of topology or bad routing)
the routing overhead grows. The DB routing and original AntNet
does not take into account the generated routing overhead
and its effect on overall network performance.
B. Introducing Load Balancing Technique to over come
the above mentioned problem
Original DB routing algorithm addresses the routing problems
but not load balancing [11]. Load balancing is heavily relied on
routing; Ant Net routing philosophy can lead to network congestion,
high delay and may create deadlock. For example a node that
lies on several routes will have a large number of packets for
different destinations in its interface queue; all these packets will
experience high queuing delays resulting in a high overall end to
end delay. Load balancing technique is needed to remove such
bottlenecks. The optimal solution found in the first phase is then
finalized by a deterministic procedure that adjusts flows in order
to achieve the precise balance of the input-output flow at each
node. A outline of this procedure is shown as follows
_______________________________
Input: Original flows stored
output: Modified flows that satisfy the condition in which each
inner node of the network, the sum of its output flows is less
than or equal to the sum of its input flows.
_______________________________
1.1 Assign starting flows to network edges
1.2 Initialize the list of nodes expected to be processed N=P
2 do
3 Take the first node P from the list N
4.1 Recalculate output flows of node P
4.2 Add all nodes that have been affected by this action to
list N
5 while(S 6= {})
Modified DB algorithm for direct representation.
For any node if its total input flow is bigger than its total output
flow, it finds a path from the given node to the source node, and
then decreases the flow along that path as much as possible.
This might be repeated several times for each node, until the
excess input flow has been completely removed.
______________________________________
input: Flows that already satisfy the condition that at each
inner node the sum of its output flows is at most as big as the
sum of its input flows.
output: Balanced flows for which the balance condition holds
is that every inner node has its total output flow equal to its
total input flow.
______________________________________
1 while (exists node v with unbalanced in-out flows)
2 flow difference= v. in - v. out
3 do
4.1 Find an acyclic path P from v to the source node s such that
all edges in the path are assigned a positive flow.
4.2 Set minimize to the flow of the edge with the minimal flow
along the path P
4.3 minimize = min(minimize, flow diff)
5 Decrease flows of all edges of path P by value of minimize
6 flow diff = flow diff - minimize
7 while (flow diff > 0)
Algorithm for finding balanced flows in every inner node of the
network using
IV. Performance Of MDB Routing Algorithm On 4x4 Grid
Topology
In this section 3 different test runs were conducted on 4x4 grid
topology with 16 nodes, of different packet loads of 3500,5000
and 10000 and the results were displayed in the following Fig.
2 to 4 and tables 1 to 3.
The study effects on increase in packet load which increases
the average time for the packet transmission in milliseconds
on each topology this is obvious because of the increase in
number of packets and number of nodes in the network.
84
I
nternatIonal
J
ournal

of
C
omputer
S
CIenCe

and
t
eChnology
IJCSt V
ol
. 2, I
SSue
1, m
arCh
2011
I S S N : 2 2 2 9 - 4 3 3 3 ( Pr i nt ) | I S S N : 0 9 7 6 - 8 4 9 1 ( Onl i ne )
www.i j c s t.c o m
Experiment 1 :
MDB routing algorithm was executed on a 16 nodes grid with
packet load as 3500. The paths generated and the results
generated were placed in the Fig. and table1
Fig. 2 : Paths generated from the MDB routing algorithm from
experiment 1
Table 1: Results from the Experiment1
Number of packets 3500
Dead 494
TTL 0
INPUT 491
OUPUT 0
NPRDies 3
Average time for packet 306.53
Experiment 2 :
MDB routing algorithm on 4x4 grid with packet traffic on the
network as 5000. The paths generated are in the Fig.and the
results are displayed in the table 2
Fig.3 : Paths generated from the MDB routing algorithm from
experiment 2
Table 2 : Results from the Experiment 2
Number of packets 5000
Dead 2492
TTL 0
INPUT 2492
OUPUT 0
NPRDies 0
Average time for packet 219.77
Experiment 3 :
Executing MDB on 4x4 with 10000 packets as load, the
paths generated are displayed in the Fig. 4 and the results
are displayed in the table 3.
Fig. 4 : Paths generated from the MDB routing algorithm from
experiment 3
Table 3: Results from the Experiment 3.
Number of packets 10000
Dead 4991
TTL 0
INPUT 4991
OUPUT 0
NPRDies 0
Average time for packet 199.84
V. Conclusions
The Grid networks are a special case of Communication
networks. Since they are easy to setup and maintain, and
have good scalability, GCNs are potentially a popular access
method for hospitals, hotels, and conference centers. This
paper studies routing algorithm for grid networks, using MDB
routing, which addresses load-balancing.
With the experimental results from the above tables, one can
conclude that the MDB is working effectively on the grid network
topology with fast transmission of data packets from the source
to the destination.
As the packet load is getting increased, the average time for
packet transmission speed is decreasing which increases the
through put.
References
[1] E. Bonabeau, M. Dorigo, G. Theraulaz, "From Natural to
Artificial Swarm Intelligence", Oxford University Press, in
press.
[2] R. Beckers, J. L. Deneubourg, S. Goss, "Trails and U-turns
in the selection of the shortest path, by the ant Lasius
Niger", Journal of Theoretical Biology, 159, 1992, pp.
397–415.
[3] D. Costa, A. Hertz, "Ants can colour graphs", Journal of the
Operational Research Society, 48, 1997,
[4] M. Dorigo, Optimization, "Learning and Natural Algorithms
(in Italian)", PhD thesis, Dipartimento di Elettronica e
Informatica, Politecnico diMilano, IT, 1992M. Young, The
Technical Writer’s Handbook. Mill Valley, CA: University
Science, 1989.
[5] M. Dorigo, L. M. Gambardella, "Ant colony system: A
cooperative learning approach to the traveling salesman
problem", IEEE Transactions on Evolutionary Computation,

I
nternatIonal
J
ournal

of
C
omputer
S
CIenCe

and
t
eChnology
85
I S S N : 2 2 2 9 - 4 3 3 3 ( Pr i nt ) | I S S N : 0 9 7 6 - 8 4 9 1 ( Onl i ne )
www.i j c s t.c o m
IJCSt V
ol
. 2, I
SSue
1, m
arCh
2011
IJCSt V
ol
. 1, I
SSue
2, d
eCember
2010
I S S N : 2 2 2 9 - 4 3 3 3 ( Pr i nt ) | I S S N : 0 9 7 6 - 8 4 9 1 ( Onl i ne )
1(1), 1997, pp. 53–66.
[6] M. Dorigo, V. Maniezzo, A. Colorni, "Positive feedback as a
search strategy", Technical Report 91016, Dipartimento di
Elettronica e Informatica, Politecnico di Milano,IT,1991.
[7] M. Dorigo, V. Maniezzo, A. Colorni, "The ant system:
Optimization by a colony of cooperatingagents", IEEE
Transactions on Systems, Man, and Cybernetics–PartB,
26(1), 1996, pp. 29–41.
[8] R. Schoonderwoerd, O. Holland, J. Bruten, L. Rothkrantz,
"Load balancing in telecommunications networks",
Adaptive Behavior, 5(2), 1996, pp. 169–207.
[9] Jangeun Jun, Mihail L. Sichitiu, “Fairness and QoS
in Multihop Wireless Networks”, Proceedings of IEEE
Vehicular Technology Conference (VTC), pp. 2936-2940,
2003
[10] S. Anuradha, G. RaghuRam, K.E.Sreenivasa murthy,
V.Raghunath Reddy, B.Satya narayana “Genesis of DB
Routing Algorithm in Unicasting Networks” ICGST Inter
National Journal on computer Networks and Internet
Research(CNIR) Volume 9, Issue I, July 2009, pp. 47-
55.
[11] S. Anuradha, G. Raghu Ram, K. E . Sreenivasa Murthy,
V.Raghunath Reddy, B.Satya Narayana, “ Evolution of New
Searching Technique in Unicast Routing Algorithm of Data
Networks”, Praise Worthy Prize’s International Review on
Computers and Software(IRECOS) Voume.4, No. 3 ,May
2009,pp 321-330.
G. Raghu Ram graduated from Sri Krishna
Devaraya University in the year 1994, MCA
from Osmania University in the year 1997.
He is presently Associate Professor and
Head in the Department of Master of
Computer Applications at G. Pulla Reddy
Engineering College, Kurnool, and Andhra
Pradesh, India. He presented three research
papers in national conferences. And published three papers
in International journals. His research areas include Artificial
Intelligence and Computer Networks.
S. Anuradha obtained her B.Sc and
MCA degrees from Osmania University,
Hyderabad in the year 1994 and 1997
respectively. She is currently pursuing
PhD at Sri Krishna Devaraya University,
Anantapur, India. She is presently working
as Associate Professor in the Department
of Master of Computer Applications at G.
Pulla Reddy Engineering College, Kurnool,
Andhra Pradesh, India. She presented four research papers
in International conferences. And published nine papers in
International journals. Her research areas include Computer
Networks and Routing Algorithms.
Dr.V.Raghunatha Reddy obtained hisB.
Sc.fromSri Krishnadevaraya University
Ananthapur India in 1991 and MCA from
Madurai Kamaraju University,Tamilnadu
India in 2002,M.Phil from MKU Tamilnadu
India in 2005 and PhD degree from SKU
Ananthapur India in 2009 He published
nine international Paper in reputed
Journal and attended four International
Conference. He is at present working as an Assistant Professor
in the Department of Computer Science Sri Krishna dearly
University Ananthapur Andhra Pradesh India. His areas
of Interest are Object Oriented Programming Computer
NetworksandCommunications.
Dr. K.E. Sreenivasa Murthy obtained
B.Tech and M.Tech degrees in Electronics
and Communication Engineering from
Sri Venkateswara University, Tirupati,
India in 1989 and 1992 respectively and
PhD degree from Sri Krishna Devaraya
University, Anantapur, India, in 1997. He
presented more than 25research papers
in various national and international
conferences and journals. He is at present working as principa
at SVIT,Ananthapur,India. His research interests include FPGA
and DSP applications.
M.Sarika obtained her Master of computer
applications from JNTU in the year 2008
and graduated at SKU and obtained her
B.Com.degree in the year 2005 Presently
working as Assistant Professor in MCA
dept. G.Pulla Reddy Engineering College,
Kurnool, AP,.INDIA. She published one
paper in a reputed International journal.
Her research Interests in AI and Computer Networks.
86
I
nternatIonal
J
ournal

of
C
omputer
S
CIenCe

and
t
eChnology
IJCSt V
ol
. 2, I
SSue
1, m
arCh
2011
I S S N : 2 2 2 9 - 4 3 3 3 ( Pr i nt ) | I S S N : 0 9 7 6 - 8 4 9 1 ( Onl i ne )
www.i j c s t.c o m