Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
_______________________________________________________
Some part of this chapter appears in LNCS-CCIS, Springer Berlin Heidelberg, Vol.142,
Pages: 544-546, ISBN 978-3-642-19541-9 (Print), ISSN 1865-0929, 2011
102
Chapter 7
Design of Enhanced Quality of
Service Aware Routing Protocol
(EQARP)
7.1 Introduction
The link reliability and delay are very important parameters in the case of
multimedia data transmission in MANET. On demand link reliable and
delay-aware routing protocol is designed by including QoS constraints
(link reliability and delay). As and when the route to transmit the
information is computed, those QoS metrics are also computed
automatically and accordingly the routes are updated in the routing table.
Hence, the route selected will be a link reliable and delay-aware route.
Other routes which are not QoS-aware are discarded from the routing
table. Broadcasting in the route discovery and the route maintenance of
Ad Hoc On demand Distance Vector Routing Protocol provokes a high
number of unsuccessful packet deliveries from the source nodes to the
destination nodes. Studies have been undertaken to optimize the
rebroadcast focused on the route discovery of the AODV. In this work,
Lifetime Ratio (LR) of the active route for the intermediate node is
introduced to improve packet delivery ratio. Lifetime is one of the metrics
103
for AODV that is stored in the routing table entry for intermediate nodes.
Lifetime [35] is an expiry time for an active route. It is also known as
deletion time for an invalid route. Initially, the static life time value from
AODV protocol is taken to find the route life time and based on that the
routing decisions are made. The proposed algorithm which takes static life
time is QoS-AODV. The static life time does not help where there is a
frequent change in the network topology and mobility is very high. So
there is a need for the dynamic computation of the link life time. Using
this, it is possible to store only reliable paths in the routing table. The
algorithm which computes multiple node disjoint paths, delay and link life
time dynamically, is designed i.e., EQARP (Enhanced Quality of Service
Aware Routing Protocol) and is compared with QoS-AODV. The proposed
EQARP protocol shows better performance for a highly dynamic network.
7.2 Life time in the original AODV
The life time is one of the parameters in the routing table of the original
AODV protocol as shown in the Figure 7.1.
Destination IP Address
Destination Sequence
Number
Next Hop
Hop count
Life Time
Figure 7.1: Original AODV Routing Table parameters
The route life time value is one of the most important parameters for
the design of an on demand Ad Hoc routing protocol. This parameter
determines the duration of the active path in the routing table to transmit
the packets reliably. The purpose of lifetime is to ensure that the routing
104
protocol does not attempt to discover a new route and/or delete an existing
active route within its lifetime.
Too long route lifetime may lead to retardation in updating the routing
table even though some paths are broken. This results in large routing
delays and control overhead from attempts to transmit across paths that
do not exist. Too short route lifetime may remove some active paths from
the routing table. This result in re-establishing the discovery process for
those paths again, causing large routing delay and traffic overhead due to
the new path search [135].
AODV route lifetime is either determined from the control packet, or it
is initialized to ACTIVE_ROUTE_TIMEOUT [40]. This means that from
the time the route is discovered, it is kept active up to predetermined
amount of time. In this protocol, ART (Active Route Timeout) is set to 3
seconds by default [2][27][136].
7.3 Computation of Route Life time ratio based on
static Lifetime and TTL
The formula for the Percentage Life Time Ratio (PLTR) is shown below:
PLTR =
*100 (7.1)
Life Time is the expiry time for the active route. It is also known as the
deletion time for an invalid route. TTL carries a time to live (TTL) value
that states for how many hops this message should be forwarded. Every
RREQ carries a time to live (TTL) value that specifies the number of times
this message should be re-broadcasted. This value is set to a predefined
value at the first transmission and increased at retransmissions. The
maximum value of the TTL is up to the Network diameter.
Retransmissions occur if no replies are received. TTL can be counter or
105
timestamp. Here timestamp interpretation is used. TTL in milliseconds is
calculated as:
TTL(in ms) = 2*Node_Traversal_Time*TTL(in hops) (7.2)
This value is set to a predefined value at the first transmission and
increased at retransmissions. Retransmissions occur if no replies are
received. The default value of AODV Node_Traversal_Time is 40ms. The
TTL is multiplied by two to include the time for acknowledgement
message. The LTR multiplied by 100 gives the PLTR (percentage lifetime
ratio) for a route. If the PLTR is above 50%, then the intermediate node
allows the rebroadcasting of RREQ messages. Consider a sample network
to demonstrate the PLTR, shown in Figure 7.2:
Figure 7.2: Sample network taking A as the source and F as the
destination node
In this network having a path (A→B→C→F) from source node (A) to
destination node (F) with 3 hops distance and suppose the lifetime is
150ms then, PLTR can be computed as:
PLTR =
*100
= 62.5%
In this case, the path from node A to F can be taken as a QoS path and
stored in the routing table, since the PLTR>50%.
C
A B F
D E
106
The lifetime is set to predetermined value as soon as the route is found.
TTL is computed based on the number of hops to reach to the particular
destination. The PLTR value should be greater than 50% to ensure that ,
the time in which route exist in that routing table must be at least half
more than the time required to traverse the nodes in that path.
7.3.1 Experimental Results
The performances of two on-demand routing protocols, viz. AODV and
QoS-AODV are compared using NS-2 simulation. The parameters used for
the simulation is shown in the Table 7.1.
Table 7.1: Simulation Parameters
Simulation time 200 seconds
Number of nodes 10,20,30,40,50,60,70,80,90,100
Map size 1000m X 1000m
Speed 10 m/s
Mobility Model Random Way Point
Traffic type Single CBR flow per node
Packet size 512 bytes
Number of Seeds 25
Propagation range 250m
Packet rate 10 packets/sec
Pause time 40 seconds
7.3.1.1 PDR vs. number of nodes at low mobility
Table 7.2 and Figure 7.3 show the PDR for each protocol versus number of
nodes for AODV and QoS-AODV protocols.
107
Table 7.2: PDR versus No. of nodes at low mobility
Figure 7.3: PDR versus no. of
nodes at low mobility
7.3.1.2 PDR vs. number of nodes at high mobility
Table 7.3 and Figure 7.4 show the PDR for each protocol versus No. of
nodes by fixing the pause time to 10 seconds and the node speed to 25m/s.
Table 7.3: PDR versus no. of nodes at high mobility
Figure 7.4: PDR versus no. of
nodes at high mobility
7.3.2 Analysis of Simulation Results
The performance of AODV and QoS-AODV routing protocols is compared
and analyzed using NS-2.34 simulator. The QoS parameter PDR is
measured by varying the node density and the node speed.
0
20
40
60
80
100
120
PD
R (
in %
)
No. of nodes
PDR vs. number of nodes at low mobility
AODV
QoS-AODV
0
20
40
60
80
100
120
PD
R (
in %
)
No. of nodes
PDR vs. number of nodes at high mobility
AODV
QoS-AODV
PDR in %
No. of nodes AODV QoS-AODV
10 99.5 99.9
20 99.3 99.8
30 99 99.75
40 98.75 99.5
50 98 99.3
60 96 99
70 92 98.7
80 90 98.5
90 86.5 98.2
100 83 98
PDR in %
No. of nodes AODV QoS-AODV
10 95 99
20 92 98.5
30 89 97
40 85 94.5
50 83 92
60 80 87
70 75 84
80 69 80
90 63 79
100 60 73
108
a) PDR vs. number of nodes at low mobility: It is observed that at
low mobility, the PDR of QoS-AODV is better in increasing the number
of nodes as compared to AODV. For example, at 80 nodes, the PDR of
AODV and QoS-AODV is 90% and 98.5% respectively. The reason is
due to reduction in routing overhead and keeping more active paths in
routing table.
b) PDR vs. number of nodes at high mobility: It is observed that
at high mobility, the PDR of QoS-AODV and AODV reduce rapidly. For
example, at 80 nodes, the PDR of AODV and QoS-AODV is 69% and
80% respectively.
Thus, from the experimental analysis, it is observed that the
performance of QoS-AODV routing protocol is better in terms of PDR
compared to AODV under low mobility situations. But the performance
degrades, as the mobility is increased. This is due to the static life time
enhancement to QoS-AODV.
7.4 Link life time based on Position and Direction
of movement
It is very difficult to determine the stability of the route based on the
lifetime value which is taken as a static, during route discovery process. In
Mobile Ad Hoc Networks, each node acts as a router. In a reactive
protocol, if a node does not know the QoS metrics of its neighbors it simply
broadcasts route request (RREQ) message to the neighboring nodes. Upon
receiving this RREQ packet, the neighboring nodes can get the QoS
metrics across their paths such as the position and movement information.
Using this information it is possible to make a routing decision regarding
whether the path is reliable or not.
In recent years, many routing algorithms were proposed for MANETs
that need the coordinates of nodes for routing process. In order to obtain
109
this information, Global Positioning System (GPS) can be employed
[137][138]. The stability of the routing path can be calculated using
current and future position of nodes. Therefore, the best path is
mathematically determined which need not be a shortest path. Clearly, if
the routing process is accomplished without any consideration to the
movement of nodes and the stability of routing path, the links can be
easily broken.
The proposal by Shahram Jamali et al. [50], provides link life time
extension to an AODV. Due to the dynamic topology changes in MANETs
the links can be easily broken. The link life time must be dynamically
computed and using this parameter and using this it is possible to check
the stability of the link. This work also concludes that it is important to
find and set up a route with longer life time as possible [139].
7.4.1 Calculation of Link Life time (LLT)
The link life time prediction is a method which requires that each mobile
device be equipped with a GPS receiver for obtaining the longitude and
latitude. Using this geographical information and considering the
network area, map position of each node can be determined. For
calculating the nodes direction and speed, the position information of
them should be updated continuously [50]. The proposed method makes
use of dynamic route lifetime instead of taking a fixed route lifetime as
mentioned in the previous section.
The Figure 7.5 shows the two mobile nodes A and B with present
locations (xa1, ya1) and (xb1, yb1) respectively. These two nodes are
within the radio range r. The first node is moving with a constant speed
Va at an angle (direction) of Ɵa with respect to x axis. Similarly, second
node is moving with a constant speed Vb at an angle (direction) of Ɵb with
respect to x axis. As nodes are mobile, the future locations of nodes A and
B are (xa2, ya2) and (xb2, yb2) respectively after the time duration t.
110
The future locations of mobile node A and B are calculated using the
following equations:
xa2(t) = xa1+ Va cosƟa t (7.3)
ya2(t) = ya1+ Va sinƟa t (7.4)
xb2(t) = xb1+ Vb cosƟb t (7.5)
yb2(t) = yb1+ Vb sinƟb t (7.6)
The distance S between mobile nodes A and B after time t is given by:
S= √ (7.7)
The mobile nodes A and B will be able to communicate with each other
as long as they will remain within the transmission range, r. So the
duration t = LLT, if S<=r. After solving equation 7.7 with s<=r and
considering t=LLT, we get
LLT √
(7.8)
where,
a = VacosƟa–Vb cosƟb , c = VasinƟa–Vb sinƟb
b = Xa–Xb , d = Ya–Yb
111
Figure 7.5: Mobile nodes A and B with current and future locations
7.4.2 Calculation of Route life time (RLT) based on
Position and direction of movement
Since a route consists of multiple links in series, it is said to be broken if
any single link among its links is broken, and thus, the lifetime of the
route becomes minimum lifetime of all links in this route [80]. Suppose a
route P consists of n links, the route P is said to be broken if any one of the
connections is broken because the corresponding two adjacent nodes move
out of each other’s communication range. The lifetime of route P or route
life time is expressed as the minimum value of the connections involved in
route P. Therefore, the RLT is equal to the minimum of LLTs among the
link life time LLT1, LLT2, and so on till the last link life time LLTn.
RLT = min (LLT1, LLT2,………, LLTn) (7.9)
The percentage route life time (PLTR) is computed using the following
equation:
PLTR =
*100 (7.10)
112
The intermediate node gets the PLTR value for each path, which it
stores in its routing table. If the PLTR for any path is above 50% then the
intermediate node stores such routes in its routing table. Otherwise, it
removes such routes from the routing table.
7.5 Algorithm for Route discovery in EQARP
(Enhanced QoS Aware Routing Protocol) by
dynamically computing average time stamp and
link life time
Suppose n is the number of mobile nodes and N is the set of mobile nodes,
N={N1, N2,…..,Nn}. Assume that the node Ni seeks to find a path to node Nj
and Nt receives the RREQ packet, where Ni, Nj, Ntϵ N and 1<i, j, t<n and
i ≠ j.
1. At the source node Ni :
a) Calculate the time taken for the packet to reach the
destination using the formula: [ Algorithm:6.3]
Timetaken = receive time – send time
b) Check whether the concerned node has an entry in
the Routing Table
c) If no entry found in the Routing table then create the
RREQ packet with field values set as :
Source = Ni, Destination = Nj, TTL =1
LLT =PLTR=0, Velocity = direction = Coordinate
position =0
d) Send the RREQ packet to the neighboring node Nt
and compute the parameters LLT and TTL
2. If ( the neighboring Node Nt is the destination node Nj ) then
begin
a) Receive all the paths arriving to it for wait period T
113
b) Select the paths which are node disjoint among the
list of paths
c) For the selected paths compute the RLT by taking
minimum value of LLT for each path
d) Compute the parameters PLTR and Average delay at
the destinationNode using the following equations:
avgTimeTakenByPackets=totalTimeTakenByPackets/C
Percentage Life Time Ratio=RLT/TTL * 100
e) Store the Average delay and PLTR in the Routing
table for these paths
f) Generate the RREP packet for unicasting to the source
node for all the node disjoint paths selected and the
paths with PLTR>50
g) Store the paths in the Routing table of Source node
end
3. If ( the neighboring Node Nt has a route to the destination node )
then
begin
a) Compute RLT and PLTR
b) Send the RREP packet to the Source node having
PLTR>50
end
4. If ( the neighboring Node Nt is neither the destination nor
having route to node Nj ) then
begin
a) Compute the delay, LLT.
b) If (the delay>=average delay) then
Do not broadcast RREQ from there.
Else
begin
Update the parameters of RREQ packet.
Rebroadcast the RREQ.
114
end
end
5. Stop
7.6 Flowchart for Route discovery in EQARP
The figure 7.6 and 7.7 show the route discovery process and QoS metrics
computation in EQARP respectively.
YES
NO
YES NO
YES NO
NO
YES
Figure 7.6: Flowchart for EQARP Route Discovery Process
Whether the
intermediate node is
destination?
Forward RREQ message along
with LLT and TTL
Start
If an intermediate
node has an entry?
Store that path in routing table
Delay>Avg_delay?
Stop
Copy RREQ and
Rebroadcast
Discard RREQ
Start with new
Route discovery A
Receive all the paths and
select node disjoint
paths among them
Compute QoS metrics
Avg_delay and PLTR for
the selected paths
Compute Delay
A
PLTR>50%
?
Generate the RREP for QoS paths
115
Figure 7.7: Flowchart on Creating a Routing Path along with QoS metrics
computation
TTL=1, LLT=RLT=PLTR=0.0, C=0, Total_Timetaken=0.0, AvgTime=0.0
Increment the Counter C to keep track of number of times an entry is made to the
Routing table at every node and Compute the Time taken
Total_Timetaken+=Timetaken
PLTR = (RLT/TTL_in_Seconds)*100
TTL=TTL+1
vgTime=Total_Timetaken/C
Store the Metrics PLTR and AvgTime in Routing Table
Stop
While destination is not
reached
Compute LLT
Compute the RLT as minimum of LLTs for each path at destination
TTL_in_Seconds=TTL*2*Node_traversal_Time
Start
116
7.7 Modelling the network and simulation
parameters
The NS-2.34 is used to analyze the performance of AODV and the new
protocol EQARP. In the simulations the following three network scenarios
are taken: (1) a low density network with N = 25 nodes; (2) a medium sized
network with 25<N<=80 nodes; and (3) a high density network with
80<N<=150 nodes. The mobile nodes are placed randomly within a 1000 m
x 1000 m area. Radio propagation range for each node is 250m and
channel capacity is 11 Mbps. Each node moves in this area according to
the random waypoint mobility model, with a speed of 5m/sec (low),
15m/sec (medium) and 25m/sec (high). Similarly, the pause time values
are considered to be 10sec as low pause time, 40sec as medium pause time
and 80sec as high pause time. The two metrics End-to-End packet and
PDR were used for performance study of AODV and EQARP. The Table
7.4 shows the standard values of QoS metrics viz. PDR and End-to-End
delay.
Table 7.4: Standard QoS metric values
Standard QoS metric values
QoS Metrics Low Medium High
1. PDR <=95% >=96% and <98% >=98%
2. Delay <=50ms 51ms to 150ms >150ms
7.7.1 Experimental Results
The base protocol used to compare the performance of EQARP is the QoS-
AODV. The metrics used in comparing these two protocols are PDR and
End-to-End delay.
7.7.1.1 PDR vs. number of nodes at high mobility
Table 7.5 and Figure 7.8 show the PDR for each protocol versus speed by
fixing the pause time to 10 seconds and the node speed to 25m/s.
117
Table 7.5: PDR vs. no. of nodes at high mobility
Figure 7.8: PDR vs. no. of nodes at
high mobility
7.7.1.2 Delay vs. No. of nodes at high mobility
Table 7.6 and Figure 7.9 show the Delay for each protocol versus speed by
fixing the pause time to 10 seconds and the node speed to 25m/s.
Table 7.6: Delay vs. No. of nodes at high mobility
Figure 7.9: Delay vs. no. of nodes
at high mobility
0
20
40
60
80
100
120
PD
R (
in %
)
No. of nodes
PDR vs. number of nodes
EQARP
QoS-AODV
0
50
100
150
200
250
Dela
y (
in m
s)
No. of nodes
Delay vs. number of nodes
EQARP
QoS-AODV
PDR in %
No.of
nodes
EQARP QoS-
AODV
10 99.8 99
20 99.5 98.5
30 99.2 97
40 98.5 94.5
50 98 92
60 97.6 87
70 96.8 84
80 96.2 80
90 95.8 79
100 95.2 73
Delay in ms
No.of
nodes
EQARP QoS-AODV
10 90 105
20 98 120
30 105 131
40 112 145
50 120 168
60 126 175
70 132 182
80 138 195
90 141 202
100 148 210
118
7.7.2 Analysis of Simulation Results
The performance of QoS-AODV and EQARP routing protocols is compared
and analyzed using NS-2.34 simulator. The QoS metrics Average delay
and PDR is measured.
a) PDR comparison: It is observed that the PDR of EQARP is better
than QoS-AODV in increasing the mobility. For example, at 80
nodes, the PDR of QoS-AODV and EQARP is 80% and 96.5%
respectively. The reason behind this is EQARP has less routing
overhead.
b) Delay comparison: It is observed that the delay of EQARP is also
improved at high mobility situations. For example, at 80 nodes, the
Delay of QoS-AODV and EQARP is 195ms and 138ms respectively.
The reason behind this is route discovery latency of EQARP is less
than that of QoS-AODV.
7.8 Summary
The performance of the routing protocols AODV and QoS-AODV are
compared. The QoS-AODV is designed initially by taking delay and static
life time into consideration. The performance of QoS-AODV is better than
original AODV. But at high mobility, this QoS-AODV fails to perform in
terms of QoS metrics. Next, AODV is improved by computing the link life
time dynamically across every link. This parameter is very important for a
highly dynamic network where the link break and route failure occurs
more frequently. It is observed from the experimental analysis that the
proposed protocol EQARP, works better than QoS-AODV protocol for high
density and high mobility situations. The improvement in performance
EQARP is due to the usage of dynamic life time.