Consideration with Dynamic Routing Security - International Journal ...
Consideration with Dynamic Routing Security - International Journal ...
Consideration with Dynamic Routing Security - International Journal ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Visweswara Rao,,Int.J.Computer Technology & Applications,Vol 3 (2), 592-606<br />
ISSN:2229-6093<br />
<strong>Consideration</strong> <strong>with</strong> <strong>Dynamic</strong> <strong>Routing</strong> <strong>Security</strong><br />
VISWESWARARAO BOLLA,<br />
Research Scholar, Department of computer science, Mother Theresa Institute of Science&Technology,Sankiteka<br />
nagar, Sathupally, khammam, A.P, India.<br />
E_Mail:bvissumtech.rao6@gmail.com<br />
Sri D.PAVANKUMAR<br />
Associate professor,Department of computer science, Mother Theresa Institute of Science&Technology,Sankiteka<br />
nagar, Sathupally, khammam, A.P, India.<br />
Abstract<br />
One of the major issues for data<br />
communication over wired and<br />
wireless networks is the security. the<br />
past work is on the designs of<br />
cryptography algorithms and system<br />
infrastructures. <strong>Dynamic</strong> routing<br />
algorithm called improved dynamic<br />
routing <strong>with</strong> security consideration,<br />
which is based on the concept of Zone<br />
<strong>Routing</strong> Protocol (ZRP) that could<br />
randomize delivery paths for data<br />
transmission. The algorithm is easy to<br />
implement and compatible <strong>with</strong><br />
popular routing protocols, such as the<br />
<strong>Routing</strong> Information Protocol (RIP) in<br />
wired networks and Destination-<br />
Sequenced Distance Vector (DSDV)<br />
protocol in wireless networks, <strong>with</strong>out<br />
introducing extra control messages.<br />
This algorithm is mainly proposed to<br />
improve the and to overcome the<br />
limitations existing <strong>with</strong> the present<br />
cryptographic algorithms and<br />
protocols. Although some designs like<br />
IP security, Secure Socket Layer<br />
provide essential security, E-Mail<br />
security<br />
they<br />
unavoidably introduce substantial<br />
overheads in the Gateway/Host<br />
performance and effective network<br />
bandwidths.<br />
Key words:<br />
<strong>Dynamic</strong> <strong>Routing</strong>, ZRP, DSDV, RIP,<br />
Secure socket Layer, Bellman Ford<br />
Algorithm<br />
INTRODUCTION<br />
In the past decades, various<br />
security enhanced measures have<br />
been proposed to improve the security<br />
and to control traffic of data<br />
transmission over public networks.<br />
Existing work on securityenhanced<br />
– data transmission includes<br />
the designs of cryptography<br />
algorithms and system infrastructures<br />
and security enhanced routing<br />
methods and techniques. Their<br />
common objectives are often to defeat<br />
various threats<br />
over the Internet, including<br />
eavesdropping, session hijacking, etc.<br />
Among many well-known designs for<br />
the systems based on cryptography,<br />
592
Visweswara Rao,,Int.J.Computer Technology & Applications,Vol 3 (2), 592-606<br />
ISSN:2229-6093<br />
the IP <strong>Security</strong> (IPSec) and the Secure<br />
Socket Layer (SSL) are popularly<br />
supported and implemented in many<br />
systems and platforms. Although<br />
IPSec and SSL do greatly improve the<br />
security level for data transmission,<br />
they introduce over heads, especially<br />
on gateway/host performance and<br />
effective network bandwidth which<br />
are For example, the data<br />
transmission overhead is 6 cycles/byte<br />
over an Intel Pentium II <strong>with</strong> the<br />
Linux IP stack alone, and the<br />
overhead increases to 62 cycles/byte<br />
when Advanced Encryption Standard<br />
(AES) is used for<br />
encryption/decryption for IPSec.<br />
Another alternative for security<br />
enhanced data transmissions is to<br />
dynamically route packets between<br />
each source and its destination so that<br />
the chance for system break-in, due to<br />
successful interception of consecutive<br />
packets for a session, is low. The<br />
intention of security-enhanced routing<br />
is different from the adopting of<br />
multiple paths between a source and a<br />
destination (say server to client for<br />
example) to increase the throughput<br />
of data transmission. A secure routing<br />
protocol is proposed to improve the<br />
security of end-to-end data<br />
transmission based on multiple path<br />
deliveries. The set of multiple paths<br />
between each source and its<br />
destination is determined in an online<br />
fashion, and extra control message<br />
exchanging is needed and a secure<br />
stochastic routing mechanism is<br />
proposed to improve routing security.<br />
Similar to the work proposed by an<br />
engineer a set of paths is discovered<br />
for each source and its destination in<br />
an online fashion based on message<br />
flooding. Thus, there is a need of<br />
mass of control messages. Yang and<br />
Papavassiliou explored the trading of<br />
the security level and problems <strong>with</strong><br />
the traffic dispersion. They proposed<br />
a traffic dispersion scheme which will<br />
help in reducing the probability of<br />
eavesdropped information along the<br />
used paths provided that the set of<br />
data delivery paths is discovered in<br />
advance. Although excellent research<br />
have been proposed for securityenhanced<br />
routing, many of them rely<br />
on the discovery of multiple paths<br />
either in an online or offline fashion.<br />
For<br />
those online path searching<br />
approaches, the discovery of multiple<br />
paths involves a number of control<br />
signals over the World Wide Web<br />
(Internet). On the other hand, the<br />
finding of paths in an offline fashion<br />
might not be to networks <strong>with</strong> a<br />
dynamic changing configuration. So,<br />
we will propose a dynamic routing<br />
algorithm which is very effective in<br />
providing security data delivery<br />
<strong>with</strong>out introducing any extra control<br />
messages.<br />
SECURE SOCKET LAYER<br />
In Secure Socket Layer we<br />
concern <strong>with</strong> the implementation of<br />
the client and server entities and the<br />
SSL transaction between respective<br />
client and server. This transaction<br />
593
Visweswara Rao,,Int.J.Computer Technology & Applications,Vol 3 (2), 592-606<br />
ISSN:2229-6093<br />
comprises of the authentication, key<br />
exchange and large data transfer. Our<br />
implementation ensures secure and<br />
reliable communication (message<br />
exchange) and data transfer (files)<br />
between the two entities. Now-a-days<br />
Information is one of the most<br />
valuable resources in the world.<br />
Whether it is a personal letter or an<br />
industrial secret, all information or<br />
data has a worth to someone. This<br />
considers issues of security and<br />
privacy for such information or data<br />
that is stored. It discusses the reasons<br />
for wishing to provide security for the<br />
data and the methods available for<br />
doing so. For secure transferring of<br />
information between the sockets at<br />
distant places which mainly requires<br />
security so that the message or data<br />
may not be tampered while it has been<br />
transferred. When a client and server<br />
communicate <strong>with</strong> each other, SSL<br />
ensures that the connection is private<br />
and secure by providing<br />
authentication and encryption.<br />
Authentication confirms that the<br />
server and the client are trustworthy.<br />
Encryption then creates a secure<br />
“tunnel” between the two, which<br />
prevents any unauthorized system<br />
from reading the data. SSL enabled<br />
clients (such as a Netscape or<br />
Microsoft browser) and SSL-enabled<br />
servers (such as Apache or IIS)<br />
confirm each other’s identities using<br />
digital certificates. Digital certificates<br />
are issued by trusted third parties<br />
called Certificate Authorities (or CAs)<br />
and provide information about an<br />
individual’s claimed identity, as well<br />
as their public key. By validating<br />
digital certificates both parties can<br />
ensure that an imposter has not<br />
intercepted a transmission.<br />
Bellman Ford Algorithm<br />
Bellman-Ford algorithm computes<br />
single source shortest paths in a<br />
weighted digraph. For graphs <strong>with</strong><br />
only non-negative edge weights, the<br />
faster Dijkstra's algorithm also gives<br />
solution to the problem. Thus,<br />
Bellman–Ford is used for graphs <strong>with</strong><br />
negative edge weights.<br />
Bellman–Ford’s basic structure is<br />
very similar to Dijkstra's algorithm,<br />
but instead of greedily selecting the<br />
minimum-weight node not yet<br />
processed to relax, it simply relaxes<br />
all the edges, and does this |V | − 1<br />
times, where |V | is the number of<br />
vertices in the graph. The repetitions<br />
allow minimum distances to<br />
accurately propagate throughout the<br />
graph, since, in the absence of<br />
negative cycles, the shortest path can<br />
only visit each node at most once.<br />
Unlike the greedy approach, which<br />
depends on some specific structural<br />
assumptions derived from positive<br />
weights, this straightforward approach<br />
extends to the general case.<br />
procedure BellmanFord(list vertices,<br />
list edges, vertex source)<br />
// This implementation takes in a<br />
graph, represented as lists of vertices<br />
594
Visweswara Rao,,Int.J.Computer Technology & Applications,Vol 3 (2), 592-606<br />
ISSN:2229-6093<br />
// and edges, and modifies the<br />
vertices so that their distance and<br />
// predecessor attributes store the<br />
shortest paths.<br />
// Step 1: initialize graph<br />
for each vertex v in vertices:<br />
if v is source then v.distance := 0<br />
else v.distance := infinity<br />
v.predecessor := null<br />
// Step 2: relax edges repeatedly<br />
for i from 1 to size(vertices)-1:<br />
for each edge uv in edges: // uv is<br />
the edge from u to v<br />
u := uv.source<br />
v := uv.destination<br />
if u.distance + uv.weight <<br />
v.distance:<br />
v.distance := u.distance +<br />
uv.weight<br />
v.predecessor := u<br />
// Step 3: check for negative-weight<br />
cycles<br />
for each edge uv in edges:<br />
u := uv.source<br />
v := uv.destination<br />
if u.distance + uv.weight <<br />
v.distance:<br />
error "Graph contains a<br />
negative-weight cycle"<br />
<strong>Routing</strong> Information Protocol (RIP)<br />
The <strong>Routing</strong> Information Protocol<br />
(RIP) is a dynamic routing protocol<br />
used in local and wide area networks.<br />
As such it is classified as an interior<br />
routing algorithm. It was first defined<br />
in RFC 1058 (1988). The protocol has<br />
since been extended several times,<br />
resulting in RIP Version 2 (RFC<br />
2453). Both versions are still in use<br />
today, however, they are considered<br />
technically obsolete by more<br />
advanced techniques, Open Shortest<br />
Path First (OSPF) and the OSI<br />
protocol IS-IS. RIP has also been<br />
adapted for use in IPv6 networks, a<br />
standard known as RIPng.<br />
.<br />
Destination-Sequenced Distance<br />
Vector routing<br />
Destination-Sequenced Distance-<br />
Vector <strong>Routing</strong> (DSDV) is a tabledriven<br />
routing scheme for ad hoc<br />
mobile networks based on the<br />
Bellman-Ford algorithm. It was<br />
developed by C. Perkins and<br />
P.Bhagwat in 1994. The main<br />
contribution of the algorithm was to<br />
solve the <strong>Routing</strong> Loop problem.<br />
Each entry in the routing table<br />
contains a sequence number, the<br />
sequence numbers are generally even<br />
if a link is present; else, an odd<br />
number is used. The number is<br />
generated by the destination, and the<br />
emitter needs to send out the next<br />
update <strong>with</strong> this number. <strong>Routing</strong><br />
information is distributed between<br />
nodes by sending full dumps<br />
infrequently and smaller incremental<br />
updates more frequently.<br />
Advantage:<br />
DSDV was one of the early<br />
algorithms available. It is quite<br />
suitable for creating ad hoc networks<br />
595
Visweswara Rao,,Int.J.Computer Technology & Applications,Vol 3 (2), 592-606<br />
ISSN:2229-6093<br />
<strong>with</strong> small number of nodes. Since no<br />
formal specification of this algorithm<br />
is present there is no commercial<br />
implementation of this algorithm.<br />
Many improved forms of this<br />
algorithm have been suggested.<br />
Multipath routing<br />
Current routing schemes typically<br />
focus on discovering a single<br />
"optimal" path for routing, according<br />
to some desired metric. Accordingly,<br />
traffic is always routed over a single<br />
path, which often results in substantial<br />
waste of network resources. Multipath<br />
<strong>Routing</strong> is an alternative approach<br />
that distributes the traffic among<br />
several "good paths instead of routing<br />
all traffic along a single "best" path.<br />
Equal-cost multi-path (ECMP) is a<br />
routing technique for routing packets<br />
along multiple paths of equal cost.<br />
The forwarding engine identifies<br />
paths by next-hop. When forwarding<br />
a packet the router must decide which<br />
next-hop (path) to use.<br />
Zone <strong>Routing</strong> Protocol<br />
The Zone <strong>Routing</strong> Protocol (ZRP)<br />
was introduced in 1997 by Haas and<br />
Pearlman. It is either a proactive or<br />
reactive protocol. It is a hybrid<br />
routing protocol. It combines the<br />
advantages from proactive (for<br />
example AODV) and reactive routing<br />
(OLSR). It takes the advantage of proactive<br />
discovery <strong>with</strong>in a node's local<br />
neighbourhood (Intrazone <strong>Routing</strong><br />
Protocol (IARP)), and using a reactive<br />
protocol for communication between<br />
these neighbourhoods (Interzone<br />
<strong>Routing</strong> Protocol(IERP)). The<br />
Broadcast Resolution Protocol (BRP)<br />
is responsible for the forwarding of a<br />
route request. It shown in the Fig 1.<br />
Zrp divides its network in different<br />
zones. That's the nodes local<br />
neighbourhood. Each node may be<br />
<strong>with</strong>in multiple overlapping zones,<br />
and each zone may be of a different<br />
size. The size of a zone is not<br />
determined by geographical<br />
measurement. It is given by a radius<br />
of length, where the number of hops<br />
is the perimeter of the zone. Each<br />
node has its own zone. radius=2-Hop<br />
E, D, B, J, E and H are border-nodes<br />
Fig 1 Zone <strong>Routing</strong> Protocol<br />
596
Visweswara Rao,,Int.J.Computer Technology & Applications,Vol 3 (2), 592-606<br />
ISSN:2229-6093<br />
Before constructing a zone and<br />
determine border nodes, a node needs<br />
to know about its local neighbors. A<br />
node may use the media access<br />
control (MAC) protocols to learn<br />
about its direct neighbors. It also may<br />
require a Neighbor Discovery<br />
Protocol (NDP).ZRP does not strictly<br />
specify the protocol used but allows<br />
for local independent<br />
implementations. NDP relies on the<br />
transmission of hello messages by<br />
each node. When the node for<br />
example node A gets a response from<br />
a node B which has received the<br />
"Hello"-messages, the node A notice<br />
that it has a direct point-to-point<br />
connection <strong>with</strong> that node B.<br />
PROBLEM ANALYSIS<br />
A security-enhanced dynamic<br />
routing algorithm based on distributed<br />
routing information widely supported<br />
in existing networks. In general,<br />
routing protocols over networks could<br />
be classified roughly into two kinds:<br />
distance-vector algorithms and linkstate<br />
algorithms<br />
Distance-vector algorithms rely on<br />
the exchanging of distance<br />
information among neighboring nodes<br />
for the seeking of routing paths.<br />
Examples of distance-vector based<br />
routing algorithms include RIP and<br />
DSDV. Link-state algorithms used in<br />
the Open Shortest Path First protocol<br />
[19] are for global routing in which<br />
the network topology is known by all<br />
nodes. Our goal is to propose a<br />
distance-vector-based algorithm for<br />
dynamic routing to improve the<br />
security of data transmission. Before<br />
further discussions, our problem and<br />
system model shall be defined.<br />
Notations and Data Structures<br />
We propose to rely on existing<br />
information exchanged among<br />
neighboring nodes (referred to as<br />
routers as well in this paper) for the<br />
seeking of routing paths. In ZRP .each<br />
node Ni maintains a routing table (see<br />
Table 1a) in which each entry is<br />
associated <strong>with</strong> a tuple (t,WNi,t‚<br />
Nexthop), where t, WNi, t, and Next<br />
hop denote some unique destination<br />
node, an estimated minimal cost to<br />
send a packet to t, and the next node<br />
along the minimal-cost path to the<br />
destination node, respectively. With<br />
the objective of this work in the<br />
randomization of routing paths, the<br />
routing table shown in Table 1a is<br />
extended to accommodate our<br />
security-enhanced dynamic routing<br />
algorithm. In the extended routing<br />
table (see Table 1b), we propose to<br />
associate each entry <strong>with</strong> a tuple (t,<br />
WNi,t‚CtNi,HtNi) CtNi is a set of<br />
node candidates for the next hop (note<br />
that the candidate selection will be<br />
elabo-rated in Procedure 2 of Section<br />
3.2), where one of the next hop<br />
candidates that have the minimal cost<br />
is marked. HtNi a set of tuples,<br />
records the history for packet<br />
deliveries through the node Ni to the<br />
destination node t. Each tuple (Nj,hNj<br />
597
Visweswara Rao,,Int.J.Computer Technology & Applications,Vol 3 (2), 592-606<br />
ISSN:2229-6093<br />
module was immediately modified to<br />
make it error free. The test cases used<br />
to evaluate the system are given<br />
below:<br />
1. Test Case: When a link/node failure<br />
is occurred.<br />
Expected Result : When a link/node<br />
failure is occurred, the data packets<br />
are transmitted by using the backup<br />
route.<br />
2. Test Case : When a link/node<br />
failure is recovered.<br />
Expected Result : When a link/node<br />
failure is recovered, the data packets<br />
are transmitted by using the original<br />
route.<br />
EXPERIMENTAL RESULTS<br />
The simulation of Secure Hybrid<br />
broadcast <strong>Routing</strong> Protocol was<br />
conducted in NS-allinone-2.34, on an<br />
Intel core i3 processor and 1 GB of<br />
RAM running Ubuntu10.0 Lts.<br />
A. Network topology<br />
In the proposed Network scenario,<br />
we simulated two types of field<br />
configurations: 50 nodes distributed<br />
over a 700m x 700m terrain and 50<br />
nodes over a 1200m x 1200m terrain.<br />
Node transmission range was taken to<br />
be 250m. The initial positions of the<br />
nodes were random. Node mobility<br />
was simulated according to the<br />
random waypoint mobility model, in<br />
which each node travels to a<br />
randomly selected location at a<br />
configured speed and then pauses for<br />
a configured pause time, before<br />
choosing another random location and<br />
repeating the same steps. We ran<br />
simulations for a constant node speeds<br />
of 0, 1, 5…and 10 m/s, <strong>with</strong> pause<br />
time fixed at 30 seconds.<br />
B. Simulation Results and Analysis<br />
In this section we present and<br />
analyse the observed results for each<br />
of the performance metric discussed<br />
in the previous section under the<br />
network and security setup. The<br />
resulting data were plotted using Gnu<br />
plot. Each data point in the resulting<br />
graphs is an average of simulations<br />
runs <strong>with</strong> identical configuration but<br />
different randomly generated mobility<br />
patterns. To compare the<br />
performances of the protocols, the<br />
following metrics are used. Packet<br />
delivery ratio: The ratio of the data<br />
packets successfully delivered at<br />
destination.<br />
i. Average Packet Delivery ratio<br />
Figure below shows the observed<br />
results for average packet delivery<br />
fraction for both the networks. As<br />
shown in the figure, the packet<br />
delivery fraction obtained using<br />
Hybrid <strong>Routing</strong> <strong>with</strong> <strong>Security</strong><br />
<strong>Consideration</strong> (HBRA) is above 96%<br />
in all scenarios and almost identical or<br />
higher than that obtained using ZRP.<br />
This suggests that HBRA is highly<br />
effective in discovering and<br />
maintaining routes for delivery of data<br />
packets, even <strong>with</strong> relatively high<br />
node show the experimental results<br />
of the throughput under different<br />
traffic loads<br />
601
Visweswara Rao,,Int.J.Computer Technology & Applications,Vol 3 (2), 592-606<br />
ISSN:2229-6093<br />
FIG1: E½SimPSl _ for AT&T US topology<br />
602
Visweswara Rao,,Int.J.Computer Technology & Applications,Vol 3 (2), 592-606<br />
ISSN:2229-6093<br />
FIG2: Effect of l on available paths for AT&T US and DANTE Europe topoliges<br />
.<br />
603
Visweswara Rao,,Int.J.Computer Technology & Applications,Vol 3 (2), 592-606<br />
ISSN:2229-6093<br />
Shows the experimental results of<br />
E½ SimPSlDDRA_<strong>with</strong>_Randomized<br />
Selector, DDRA <strong>with</strong>out Randomized<br />
Selector, ECRA, and SPRA under the<br />
AT&T topology. From this figure, we<br />
observe that our DDRA-based<br />
methodologies greatly outperform<br />
SPRA and ECRA for all l values<br />
under investigation,1 1hich indicates<br />
that our DDRA provides larger path<br />
variation and, thus, more secure<br />
packet routing. Also, the E½SimPSl _<br />
values for SPRA, ECRA, and<br />
DDRA_<strong>with</strong>out_Randomized<br />
Selector increase as l increases.The<br />
increasing rates for SPRA and ECRA<br />
are much larger than those for<br />
DDRA_<strong>with</strong>out_RandomizedSelector<br />
especially when l is large.<br />
Specifically, the E½SimPSl _ value<br />
for SPRA is the same as the length of<br />
minimal-cost path because all packets<br />
always go through the minimal-cost<br />
path between source-destination<br />
pairs.Onthe other hand,whenl<br />
increases, E½SimPSl _ for<br />
DDRA_<strong>with</strong>_RandomizedSelector<br />
increases and then decreases. For all l<br />
values, the performance of<br />
DDRA_<strong>with</strong>_Randomized Selector is<br />
better than that of<br />
DDRA_<strong>with</strong>out_RandomizedSelector<br />
. The RandomizedSelector can<br />
prevent from selecting the previous<br />
nexthop for the current packet<br />
delivery and therefore avoids that<br />
consecutive packets are transmitted to<br />
the same nexthop<br />
Fig. shows the impact of l on the<br />
average number ðANÞ of available<br />
paths for each source-destination pair<br />
in AT&T US and DANTE Europe<br />
topologies. The figure indicates that<br />
AN increases as increases. Also, we<br />
observe that for a fixed l, there are<br />
more available paths in the AT&T US<br />
topology that those in the DANTE<br />
Europe topology. The reason is that<br />
the average number of links between<br />
the nodes in the AT&T US topology<br />
is larger than that in the DANTE<br />
Europe topology.Thus,<br />
more nexthop candidates can be<br />
selected in the AT&T US topology<br />
than in the DANTE Europe topology<br />
while packets are transmitted over the<br />
network by using our proposed<br />
security-enhanced dynamic routing<br />
algorithm.<br />
604
Visweswara Rao,,Int.J.Computer Technology & Applications,Vol 3 (2), 592-606<br />
ISSN:2229-6093<br />
CONCLUSIONS<br />
FUTUREWORK<br />
AND<br />
References<br />
The simulation results for<br />
Hybrid broadcast <strong>Routing</strong> <strong>with</strong><br />
<strong>Security</strong> <strong>Consideration</strong> under different<br />
mobility patterns and traffic scenarios<br />
show that the proposed protocol is as<br />
efficient as ZRP in discovering and<br />
maintaining routes. However, the<br />
impact of the overhead caused is<br />
almost insignificant and negligible as<br />
compared to the proposed degree of<br />
security, which provides compared to<br />
its other counterparts.<br />
The advantages of a multipath<br />
approach are clearly exemplified. We<br />
can conclude that the multipath<br />
approach can increase confidentiality.<br />
[1]. C. Siva Ram Murthy and B. S<br />
Manoj, “AdHoc Wireless Networks,<br />
Architecture and Protocols”, Prentice<br />
Hall PTR, 2004.<br />
[2]. Stefano Basagni, Macro Conti,<br />
Silvia Giordano andIvan Stojmenovic,<br />
“Mobile Ad Hoc Networks”, IEEE<br />
press, A john Wily & Sons, INC.<br />
publication, 2003.<br />
[3]. Haas Z. J., Pearlman M. R., and<br />
Samar P., “The Zone <strong>Routing</strong><br />
Protocol(ZRP)”,IETF Internet Draft,<br />
draft-ietf-manet-zone-zrp-04.txt, July<br />
2002.<br />
[4]. Jan Schaumann, “Analysis of<br />
Zone <strong>Routing</strong> Protocol”, Course<br />
CS765, Stevens Institute of<br />
Technology Hoboken, New Jersey,<br />
USA, 8th December 2002.<br />
[5]. George Aggelou,“Mobile Ad Hoc<br />
Networks”, 2nd edition, Mc GRAW<br />
Hill professional engineering,2004.<br />
[6]. T.H. Cormen, C.E. Leiserson, and<br />
R.L. Rivest, Introduction to<br />
Algorithms. MIT Press, 1990.<br />
605
Visweswara Rao,,Int.J.Computer Technology & Applications,Vol 3 (2), 592-606<br />
ISSN:2229-6093<br />
[7]. I. Gojmerac, T. Ziegler, F.<br />
Ricciato, and P. Reichl, “Adaptive<br />
Multipath <strong>Routing</strong> for <strong>Dynamic</strong><br />
Traffic<br />
[8].willam stallings cryptography and<br />
network security fourth edition<br />
[9]. Bellman, Richard (1958), "On a<br />
routing problem", Quarterly of<br />
Applied Mathematics 16: 87–90,<br />
MR0102435.<br />
606