Content uploaded by Khaled Elleithy
Author content
All content in this area was uploaded by Khaled Elleithy
Content may be subject to copyright.
Content uploaded by Khaled Elleithy
Author content
All content in this area was uploaded by Khaled Elleithy
Content may be subject to copyright.
Content uploaded by Khaled Elleithy
Author content
All content in this area was uploaded by Khaled Elleithy
Content may be subject to copyright.
A Novel Optimization of the Distance Source
Routing (DSR) Protocol for the Mobile Ad Hoc
Networks (MANET)
Syed S. Rizvi
1
, Majid A. Jafri, and Khaled Elleithy
Computer Science and Engineering Department
University of Bridgeport
Bridgeport, CT 06601
{srizvi, majidals, elleithy}@bridgeport.edu
Aasia Riasat
Department of Computer Science
Institute of Business Management
Karachi, Pakistan 78100
aasia.riasat@iobm.edu.pk
1
Contact author: srizvi@bridgeport.edu,
Abstract- This paper presents a new scheme for the Distance
Source Routing (DSR) protocol which shows the improvement
over the two major metrics of the DSR protocol: Route
Discovery and Route Maintenance. In addition, we present a
mathematical model that includes probability density function
for these two observed metrics. Our simulation results
demonstrate a significant improvement in the route discovery,
transmission time, and the overall network utilization. As an
interesting side result, our analysis also shows that the
proposed model can be used to effectively reduce the packet
losses.
Keywords- DS-CDMA, bit error rate, data throughput, multiuser
communications
I. INTRODUCTION
The Dynamic Source Routing (DSR) protocol is dealt
under On-Demand Routing (ODR) protocol which is just an
exact opposite to the Table-Driven Routing (TDR) [2, 3].
Generally, there are two main phases use in the DSR
protocol. One is the Route Discovery (RD) phase which
discovers all the possible paths for the packets to be
transferred from a particular source to a destination. It is
essential to properly maintain the RD phase since
maintaining a separate table for storing routing details
involves cost issues. The second phase of the DSP protocol is
the Route Maintenance (RM) phase which fixes all the
possible paths from one particular source to a destination [5].
In DSR, the packets are transmitted only one time for each
node. If the node does not receive the packet, the previous
node is responsible to make attempts in order to transmit the
packet. On the other hand, if the destination node receives
the packet successfully, an acknowledgment is transmitted
back to the source node for the received packet. Since the use
of the DSR protocol does not require the maintenance of a
cache table, it allows us to avoid unnecessary updating works
which results space and time saving advantages.
In the existing DSR scheme, the malfunctioning of one or
more links along a certain route requires the retransmission
of all packets back to the originating source node. This
unnecessary amount of retransmission results a significant
transmission overhead that can severely degrade the overall
network performance by increasing the average time delay.
In order to minimize the transmission overhead and
maximize the network throughput, we present an alternative
scheme that can be used to optimize the performance of DSR
protocol. Specifically, our proposed scheme suggests
improvement in the RD and the RM metrics of the DSR
protocol. Based on the proposed optimization, we derive a
mathematical model which proves the correctness of the
proposed scheme.
II. PROPOSED OPTIMIZATION FOR THE DSR PROTOCOL
Our main goal is to maintain the original underlying
architecture of the DSR protocol. Therefore, we consider the
DSR scheme as a black box. The DSR protocol fails to
maintain route consistency in the presence of broken links.
When one of the links goes down, the DSR protocol locates
an alternate route and transmits back the packet to the source
node where the packet was originated. On contrary to the
actual scheme of the DSR protocol, our proposed scheme
uses a reserve direction search method. In our proposed
scheme, the packets would be transmitted to the immediate
prior node where the actual error was occurred. The
proposed scheme then finds one or more alternative routes
from the current location to the destination. This implies that
the whole searching procedure of the proposed scheme will
be done in the opposite direction starting from the
destination node. Our simulation results demonstrate that the
proposed scheme considerably increases the chance of
finding a valid route for salvage packets that are typically
stored in the send buffer.
For instance, consider an example for locating a route
based on the reverse direction search scheme as shown in
Fig. 1. It can be observed that the route finds by the RD
procedure from node A (source node) to L would be:
ADEIL. During transmission of the packets, it is
detected at run time that the shortest link between node E
and I goes down. Consequently, the proposed scheme
immediately starts searching the best available alternate
routes. In order to reach the destination node, the proposed
scheme locates the neighboring nodes (i.e., node B, D, and H
from node E). This process of finding the alternate route
from the location of error results an optimal alternate route:
A
D
E
I
H
L. This implies that our proposed scheme
neither send any feedback to the destination node A nor it
initiates the route discovery from the source point. Therefore,
repeating this search in the reverse direction from the current
location of error to the neighboring nodes results a
significant increase in the chance of finding a valid
optimized route.
A. Proposed Reverse Direction Search Scheme
In order to formulate the proposed scheme, we present a
model that shows simple steps that need to be implemented
for finding a valid and optimize route in the presence of link
failures. The model is presented in Fig. 2. The model is
typically divided into two parts. The upper part of the model
represents the RD procedure where as the lower part
represents the RM procedure. The RD procedure is based on
an exhaustive search of an internal cache. During the
transmission of a packet, if one of the links goes down, the
proposed scheme mentions that the packet will be
immediately forwarded to the next available node and starts
transmitting from the new location. Unlike the DSR
protocol, the proposed scheme minimizes the transmission
overhead by avoiding the unnecessary transmission of data to
the source node in the presence of a faulty link. In other
words, the proposed scheme does not provide any feedback to
the source node that leads to a significant improvement in
the network throughput. Since the RD can be done on the
current node, we do not need to focus on the source node.
This implies that the proposed scheme suggests the best
Fig. 2. Flow chart showing proposed model of DSR algorithm
Fig.1. Finding the alternate path in DSR protocol according to the
proposed scheme
delivery of the packets even in the presence of link failure. In
addition, the repetition of the packets due to the flooding will
be cut down.
In the proposed model, we mainly focus on the RD and the
RM. During the RD process, if the entries are found in the
internal cache of the next node, the proposed scheme
determines the optimal path that will be used to forward all
the packets to the next node. At that current node location,
the same procedure for searching the optimal path will be
repeated over the passage of time in order to find the best
path towards the destination. An empty entry in the internal
cache represents that there is no valid route exist for a
particular destination. In such a scenario, the proposed
scheme will lookup into the next neighbor’s cache and
determine the best available route for the desired destination.
Once the optimal route is discovered, the packet can then be
transmitted. In the RM process, whenever there is a link
failure along the path, the packet would not go further at the
point of error and there is no need to send any feedback to
the original source node. Instead, the proposed scheme
determines and performs the RM process on the best
available alternate path.
B. Mathematical Model
We derive our mathematical model based on the proposed
reverse direction scheme. In our mathematical model, we
show that the transmission of packets via an alternate route
is more efficient as compared to transmitting packets from
the source node using a primary route. This is especially true
in the presence of error. All system variables, along with
their definition, are listed in Table I.
The accuracy of the proposed scheme is essentially
dependent on how efficiently we can discover the alternate
routes in the presence of faulty links. In general, the accuracy
is partially related to a certain interval by which we perform
the RD procedure for a specific type of network traffic such
as a stream of packets. In particular, we first need to derive
an expression for a random variable, x, that can be used to
characterize the behavior of RD process with respect to time.
Therefore, in order to implement the proposed scheme, one
must measure the frequency of route discoveries. In order to
determine the interval between the route discoveries, the
following mathematical expression can be derived for a
random variable, x:
( )
xf x dx
+∞
−∞
∫
(1)
It should be noted that equation (1) is based on the PDF
which is used to find the frequency of route discovery for a
particular pair of source and destination.
Figure 4 represents the proposed scheme with the primary
and the secondary paths along with their corresponding
links. It can be seen in Fig. 3 that the node P represents the
primary route whereas the node S represents the secondary
route. If an error occurs in the primary route, the proposed
scheme will immediately discover an alternate route S
1
rather
than going back to the source node A. In other words, in the
presence of faulty links, the proposed scheme searches the
internal cache and determines the alternative route S
1
which
is typically stored in the local cache.
For this particular scenario, the success of the proposed
scheme is heavily dependent on the rate at which one may
need to execute the RD procedure. In addition, the success of
the proposed scheme is not only dependent on the rate at
which the RD procedure will be performed but also
dependent on the accuracy and the efficiency by which the
alternate routes will be determined. In order to find the
frequency of an alternative RD, we assume that an event E
might occur at a discrete point in time in the network which
causes an error in one of the two types of routes (i.e., the
primary P and the secondary S routes). Thus the
transmission of an event can be mathematically described as:
1 1 2 1 2 1 3 1 2 3 2 1
( ( ) ) ( ( ( ) ) ...
E PS P P S S P P P S S S
= + + + + +
(2)
TABLE I
SYSTEM PARAMETERS AND DEFINITIONS
Parameters Description
P
i
This represents the ith link in a primary path.
S
i
This represents the ith link in a secondary path.
X
Pi
Life time of the ith primary route.
X
S
i
Life time of the ith secondary route.
X
R
Minimum life time for the collection of all values in
the primary path links
T
Intervals for route discovery
E
o
An event that shows any of the given link fails
f
T
(t)
Frequency of route discovery
Z
i
Maximum life time among all available values.
i
P
Represents the faulty primary link due to an event E
i
S
Represents the faulty secondary link due to an event E
Fig. 3: Proposed scheme with primary and secondary path and their links
where
i
P
represents the faulty primary-link where as
i
S
represents the faulty secondary-link which caused due to the
occurrence of an event E at discrete point in time within a
network.
Equation (2) represents a generic equation that shows how
the occurrence of an event in the network may cause an error
in the alternate routes. Equation (2) can be further extended
for the maximum K number of forwarding links within the
available primary paths. It should be noted that the
occurrence of an event E is representing a cause of
malfunctioning in the currently used valid route. Taking
these factors into account, one may write the following
mathematical expression:
1 1 2 2 1 3 3 2 1
1 1
...
k k k
E P S P S S P S S S
P S S S
−
= + +
+ +
(3)
where
i
P
and
i
S
in (3) represent the faulty primary and
secondary links, respectively. Both of these faulty links are
caused due to the occurrence of an event E at a discrete point
in time within a network. It should be noted that we only
consider the values of the most forwarding links that one
may find within the primary path links from the generic
equation (2).
One of our observations about the two phases of the
proposed scheme is the life time of the primary path which
we use to transmit the packets to the desired destination in
the presence of the faulty links. In other words, in order to
effectively implement the proposed scheme, we must
determine the minimum value of the life time for primary
path links. This calculation is essential, since the ration of
determining the accurate valid primary links is critically
dependent on the knowledge of accurate values of lifetime.
The minimum life time of primary path links is simply
chosen from one of the primary links that has a smallest
value for the life time. In other words, if one of the ith
primary routes has the smallest life time value, this will be
chosen as a minimum life time value for the primary path
links. This hypothesis can be changed into a simple
expression:
1 2
, ,....,
R p p pk
X Min X X X
= (4)
where X
R
in (4) represents the minimum life time value for
the collection of all values in the primary path links. The
right hand side expression of (4) represents the life time of
each individual primary route starting from X
p1
to
X
pk
. These
values are considered as a life time of the sub links in the
primary path. Similar to (4), we can further extend our
mathematical model for computing the interval of time for
the RD procedure:
1 2
, ,....,
s s sn
T Max X X X
= (5)
where T represents the intervals of time for the RD and X
Si
represents the life time of the ith secondary route.
Equation (5) gives an estimate of the time to be taken by
the proposed scheme for the RD procedure. This value is
evaluated from the maximum values of the collected time in
the sub links of the secondary path. The right hand side
expression of (5) represents the life time of each individual
secondary route starting from X
s1
to X
sn
. For the sake of the
simulation and the performance evaluation, we assume that
the value of T will be measured in millisecond. Combining
(4) with (5), we can compute the value of the alternative
route discovery as follows:
1 1 2 2 1
1 1
( . ), ( . . )
... ( , , ,.... )
p s p s s
pk sk sk s
Max X X Max X X X
T Min
Max X X X X
−
=
(6)
Equation (6) gives the value of the alternative RD. This
can be considered as the optimum value which is determined
from all the available maximum values for both the primary
and the secondary links. Using (6), we can compute the
values for the RD metrics which is one of the subparts of the
proposed scheme.
Z
i
=
1 1
, , , ....,
p i s s i s
M a x X X X X
−
(7)
where Z
i
represents the maximum life time among all
available values for both primary and the secondary paths.
Recall (1), we can now derive an expression for the
frequency of RD using equations (2) to (6).
1
1
( ) (1 )
i k
NN
t t
T i
i
k k i
f t e e
λ λ
λ
− −
=
= ≠
= −
∑ ∏
(8)
where the right hand side of (8) represents the frequency of
RD.
Equation (7) also has a significant impact on the RD for
the alternate path. Implementing the results of (7) on (8), we
can derive a new expression for the frequency of the RD
which take into account the maximum life time among all
available values for both primary and the secondary paths. In
addition, this implementation describes the PDF in Z
i
with
respect to the RD metrics.
(
)
( )
1
1
( )
1
1,
( )
( )
(1 )
i
j
i
i
Zi
i
j
k k j
i e j t
f t
e k t
λ
λ
λ
−
+
+
−
=
= ≠
=
−
∑
∏
(9)
where
( )
/ 1, 2....
i
j
ki l for j i
λ
= → = and for 1/l
j=i+1.
Equation (9) describes the summation of all the possible
routes which can lead us to the desired destination. Equation
(9) can be further extended for the following given
expressions:
1 1 2 2 1
1 1
( . ), ( . . )
... ( , , ,.... )
p s p s s
pk sk sk s
Max X X Max X X X
T Min
Max X X X X
−
=
1 1
( , , .... )
i pi si si s
Z Max X X X X
−
=
1 2 3
( , , .... )
k
T Min Z Z Z Z
=
Based on the above three expressions, we can approximate
the PDF of T for the frequency of RD as follows:
0
( ) lim [ ]/
T
dt
f t p t T t dt dt
→
= ≤ ≤ + (10)
Equation (10) gives the value for the frequency of the RD
in terms of a PDF function. Relating (8) and (9) with (10),
we can derive the following mathematical expression
1 1, 1
1
1, 1
( ) ( ) [ ]
( ) ( ) (1 ( ))
kk
T Zi j i
i j j
kk
T Zi i
i
j j
f t f t p z z
f t f t Fz t
= = ≠
=
= ≠
= >
= −
∑ ∏
∑
∏
(11)
where, F
zi(t)
in (11) was introduced from (7) to make Z
i
as a
function of PDF.
Equation (11) shows that we derived the expected
expression which can be used to compute the interval
between the rout discoveries. In other words, one could use
(11) to determine the frequency of the alternate RD process.
The same frequency value can be used to measure the
efficiency of the network. In addition, the final results show
that the use of the proposed reverse direction scheme with
the derived mathematical model can effectively minimize the
transmission delay especially in the presence of collisions
(links error) or faulty links due to the malfunctioning.
III. SIMULATION RESULTS
We simulate our model based on the predicted data from
the existing DSR model suggested in [1, 4]. For the sake of
simulation and the performance evaluation, we consider two
major metrics for RD and RM. These metrics are considered
for the evaluation of the efficiency of a network.
For the sake of the first simulation (see Fig. 4), we
characterize the behavior of the RD phase of the proposed
scheme with respect to the number of nodes present in the
network. The purpose of this experiment is to show the
performance of the RD phase for discovering the alternate
primary and the secondary path. During the simulation, we
consider that as the number of nodes increases in the
network, the more packets will be accumulated in the
network that could affect the performance of the RD phase. It
can be clearly evident in Fig. 4 that the RD phase of the
proposed scheme performs better for the primary paths
discoveries than for the secondary path. When we have small
Fig.4. Number of nodes versus RD
Fig. 5. Packet loss in fractions versus number of nodes
number of nodes, it can be seen in Fig. 4 that the
performance of the RD phase for both primary and secondary
path discoveries is overlapping. However, as network grows
in terms of the number of nodes, the performance differences
between the primary and the secondary path is obvious.
Fig. 5 shows the packet losses (in the fraction value) with
respect to the number of nodes during the transmission using
both primary and the secondary paths. In addition, Fig. 6
represents a comparison between the time delay (represents
in seconds) and the number of nodes. It can be seen in Fig. 6
that the time required to discover the primary paths using the
RD phase is very low as compared to the time required to
discover the secondary paths.
Based on the simulation results of Fig. 6, we can observe
that the time delay for primary paths is not only small but
also linear with respect to the number of nodes. In other
words, when we increase the number of nodes in the
network, more packets will be accumulated that make a
linear increase in the time delay for discovering the
secondary paths which is not really desirable as far as the
optimum performance of the DSR protocol is concerned.
IV. CONCLUSION
In this paper, we presented a new scheme that improves
the retransmission mechanism for the existing DSR protocol.
In order to support our hypothesis, we provided a complete
mathematical model that shows the formulation of the
proposed scheme. In particular, we investigated the RD and
the RM phases with respect to the proposed reverse direction
scheme. We also showed that how effective the proposed
scheme would be when we implement it with the reverse
direction search for discovering the primary paths. Our
analysis also suggested that the discovery of alternate
primary paths from the current source of error significantly
improves the network performance in terms of RD process,
time delay, and the packet losses. Moreover, we have
experimentally verified that both the RD and the RM metrics
perform well with the proposed scheme than the existing
infrastructure of the DSR protocol. Our performance
evaluation is also well supported by the simulation results
presented in this paper.
REFERENCES
[1] P. Papadimitratos and Z. Haas, “Secure Routing for Mobile Ad hoc
Networks,” In Proceedings of the SCS Communication Networks and
Distributed Systems Modeling and Simulation Conference (CNDS
2002), San Antonio, TX, January 27-31, 2002.
[2] J. Raju and J. Garcia-Luna-Aceves, “A comparison of on demand and
table driven routing for ad-hoc wireless networks,” In Proc. IEEE
International Conference on Communications (ICC 2000), June 2000.
Volume 3, Issue 2000, pp. 1702 – 1706, 2000.
[3] J. Raju and J. Garcia-Luna-Aceves, “Efficient On-Demand Routing Using
Source-Tracing in Wireless Networks,” In Proc IEEE Global
Telecommunications (GLOBECOM 2000), Vol. 1, Issue 2000, pp. 577 –
581, November 2000.
[4] B. Johnson, A. Maltz, and Y. Chun, "The Dynamic Source Routing
Protocol for Mobile Ad Hoc Networks. (DSR)," IETF INTERNET
DRAFT, 24 February 2003.
[5] V. Park and M. Corson, "A Highly Adaptive Distributed Routing
Algorithm for Mobile Wireless Networks," Sixteenth Annual Joint
Conference of the IEEE Computer and Communications Societies.
Driving the Information Revolution (INFOCOM '97), pp.1405, 1997.
Fig. 6. Time delay versus number of nodes
