The basic feature of DSR protocol is that the route cache of each node stores detailed information of each route from source to destination. Thus each node can

determine traffic load from it distributed to all connections in network based on

routing information in its route cache. Thence, when source node receives RREP

for route discovery results, based on routing information in its route cache, the

source node can select a route so that traffic load distributes to all connections is

most balanced. This is the idea of selecting load balancing route of SLBQT-DSR.

27 trang |

Chia sẻ: honganh20 | Ngày: 14/03/2022 | Lượt xem: 285 | Lượt tải: 1
Bạn đang xem trước 20 trang tài liệu **Improve the performance of mobile ad hoc network using load balancing routing technology ensuring quality of transmission**, để xem tài liệu hoàn chỉnh bạn click vào nút DOWNLOAD ở trên

For the topology as shown in Figure 2.16, from A to H can use the route A→ E
6
→ G→ I→ H. Although hopcount of this route is 4, SNR of that is 24.1 dB. This
value is better than SNR of the route A→ E→ C→ H that AODV found.
2.4. QoT of the routes when using load balancing routing protocols
2.4.1. The principle of load balancing routing technique
G
C
I
B
F
A
E
D
H
31
32
28
32
32
29
31
29
29
24
24
3532
28
RREQ is continued to broadcast
RREQ is discarded
RREP is replied to source node
Figure 2.17. An example of load
balancing routing in MANET network
Load balancing routing is the routing tech-
nique in which the route selection crite-
rion is the uniform load traffic distribution
across all connections in the network.
2.4.2. QoT of the routes
Consider an axample of the route discov-
ery as shown in Figure 2.17 with the FMLB
load balancing routing algorithm [70] used,
K is set to 3. Considering case A wants to
transmit data to H. According to the princi-
ple of route discovery by broadcasting the
RREQ packets, three routes found are A→
E → C → H, A → E → G → I → H and
A → B → D → H. SNR of the routes are
23.86, 24.04 and 20.2 dB, respectively. Thus, only the second route satisfies QoT
constraint. Meanwhile, all three routes are used. Therefore, data packets are trans-
mitted on the first route and the third route with non-guaranteed QoT.
2.5. Evaluate QoT and network performance using simulation method
2.5.1. Simulation scenarios
To evaluate QoT of the data transmission routes and its effect on the MANET
performance, the author has simulated based on OMNeT++ [10].
Table 2.5. Simulation parameters
Parameters Setting Parameters Setting
Network Size 1000m × 1000m BER threshold 10−6
Modulation format 256-QAM Required SNR 23.5 dB
MAC protocol 802.11ac Noise model Thermal noise
Number of nodes From 20 to 50 Temperature 3000K
Transmit Power 19.5 dBm Transmission Range 250 m
Receiver Sensitivity -68 dBm Speed of nodes 5 - 20 m/s
2.5.2. Simulation results of DSR protocol
The result in Figure 2.19 shows the SNR at the receiver of the destination node.
There are many routes that does not satify the constraint of QoT since its SNR is
less than required SNR. This is the cause of the increasing BPD in the network.
7
Figure 2.19. SNR of the routes in case of
DSR protocol
21.00
22.00
23.00
24.00
25.00
26.00
20 25 30 35 40 45 50
DSR
21.00
22.00
23.00
24.00
25.00
26.00
20 25 30 35 40 45 50
Tổng số nút mạng
S N
R
n h
ỏ
n h
ấ t
( d
B )
Giá trị yêu cầu
DSR
QTA-DSR
M i
n i m
u m
S
N R
( d
B )
Network size (nodes)
Required SNR
Figure 2.21.Minimum SNR in case
of DSR protocol
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
B P
D
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Tải lưu lượng (Erlang)
B P
D
BPD toàn phần
BPD do QoT
Traffic load (Erlang)
BPD overall
BPD due to QoT
Figure 2.24. BPD versus traffic load
in case of DSR protocol
The existence of many routes that do not sat-
isfy QoT constraint has increased BPD as
shown in Figure 2.24. BPD due to QoT is not
satisfied to account for nearly 50% of the to-
tal BPD.
2.5.3. Simulation results of AODV
For AODV, SNR of the routes as shown in
Figure 2.29. There are many routes that does
not satify the constraint of QoT (is less than
23.5 dB). This is the cause of increasing
BPD, this is clearly visible from Figure 2.31.
Figure 2.29. SNR of routes in case of
AODV protocol
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
5 10 15 20
Tốc độ di chuyển (m/s)
B P
D
BPD toàn phần
BPD do QoT
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
5 10 15 20
B P
D
Mobility speed (m/s)
BPD overall
BPD due to QoT
Figure 2.31. BPD versus mobility
speed of AODV protocol
2.6. Conclusion of chapter 2
Chapter 2 presents the research results about the physical effects happening on
the data transmission routes and its impact on MANET network performance. The
simulation results have proved that, these effects is the cause of BPD increase, lead-
ing to the reduction of network performance. Therefore, it is essential to improve
routing algorithms to ensure QoT and improve network performance.
8
CHAPTER 3
LOAD BALANCING ROUTING ENSURING QUALITY OF TRANSMISSION
BASED ON TRAFFIC LOAD OFFERS TO EACH ROUTE
3.1. Introduction
The research results in Chapter 2 have shown that, load balancing routing can be re-
solved traffic bottleneck in the network. However, it can decrease QoT because the
routes may pass through multiple hops. To ensure the QoT of the data transmission
routes, several works have proposed routing algorithms that take into account the
constraints of some QoT [5, 24, 46, 58], where the proposed algorithms attempt
to find out the best QoT route. This therefore improves the QoT in the network.
However, for the mesh topologies such as MANET, the routing technique with the
best QoT can increase the bottlenecks due to unbalanced traffic load.
Thus, one problem to consider is how to combine harmony between QoT constraint
routing and load balancing routing, to find a set of routes that load traffic distribute
balancedly for all links, while satisfying the constraint of QoT as shown in Figure
3.2. For this idea, the author proposes a load balancing routing algorithm, while en-
suring QoT of the routes. The load balancing route is chosen based on information
about probability of blocking packets from source to destination. The proposed
algorithm is called LBRQT (Load Balancing Routing ensuring QoT).
Shortest path
or best QoT
routing
Traffic load distributes
unbalancedly to all
connections
Bottlenecks
There are some long
routes (pass through
multiple hops)
Decreasing QoT
Load balancing
routing under
constrain of QoT
Load balancing
routing
Figure 3.2. The idea of proposing load balancing routing under QoT constraints
3.2. Relevant theory
3.2.1. Analyze the blocking probability of data packet using queue theory
Considering a hop j (hi j), assume that the data packet arrivals follow Poisson dis-
tributions, the packet transmission times are exponentially distributed. Thus hi j is
modeled as M/M/1/L queuing [6, 63]. By solving the steady-state balance equation,
we determine BPD on hi j as follows:
B(h)i j =
ρLi j(1−ρi j)
1−ρL+1i j
if ρi j 6= 1
1
L+1 if ρi j = 1
(3.4)
9
where λi j and µi j are the arrival and service rates of data packets, ρi j = λi j/µi j is
traffic density distributed to hi j . Let B
(r)
sd is BPD of route rsd , we have
B(r)sd = 1− ∏
∀hi j∈rsd
(1−B(h)i j ) (3.7)
3.2.2. Analyze end-to-end delay based on the queue theory
End-to-end delay (EED) of a route is determined by:
τ(r)sd = ∑
∀hi j∈rsd
τ(h)i j (3.9)
where τ(h)i j is delay of hi j , consists of four components which is the processing
delay (τ(i)p ), queuing delay (τ
(i)
q ), transmission delay (τ
(i j)
t ) and radio propagation
delay (τ(i j)r ) [18]. Since τ
(i)
p and τ
(i j)
r are small enough, they are able to ignore,
τ(i j)t is determined based on the bit rate of the channel and data packet size, τ
(i)
q is
determined based on the queue mechanism at the network nodes. As analyzed in
Section 3.2.1, M/M/1/L queue mechanism is used, so τ(i)q is determined by [19]:
τ(i)q =
L
λi j(1−B(h)i j )
+
1
µi j
(3.11)
where L is the average length of the queue, determined by [19].
3.3. The idea of the proposed algorithm
3.3.1. Analytical model
The idea of proposing LBRQT algorithm is to combine balancing routing and QoT
constraint routing. To implement this idea, the objective function is to minimize
BPD on each route. The constraint are defined including QoT and EED.
In order to formulate LBRQT routing algorithm, the author defines a matrix Xsd =[
x(sd)i j
]
n×n which is the matrix denoting the links of the route rsd , where each ele-
ment x(sd)i j is determined by
x(sd)i j =
{
1 if rsd passes through ci j
0 otherwise
(3.12)
Therefore, the equation (3.7) is denoted according to x(sd)i j as follows:
B(r)sd = 1−
n
∏
i=1
n
∏
j=1
(1− x(sd)i j B(h)i j ) (3.13)
10
Thence, LBRQT algorithm is modeled to nonlinear programming problem:
Miniminze (B(r)sd ) (3.19)
Subject to the following constraints due to:
∑
i∈N
x(sd)i j − ∑
k∈N
x(sd)jk =
−1 if j = s
1 if j = d
0 otherwise
(3.20)
N
∑
i=1
N
∑
j=1
(
x(sd)i j τ
(h)
i j
)≤ τth (3.21)
N
∑
i=1
N
∑
j=1
( 1
β (h)i j x
(sd)
i j
)
≤ 1
βreq
if AF is used
min
x(sd)i j =1
(
β (h)i j
)
≥ βreq otherwise
(3.22)
(x(sd)i j −1)x(sd)i j = 0 (3.23)
The constraint conditions of (3.20), (3.21), (3.22) and (3.23) are the flow conser-
vation, EED delay, QoT and integer constraints, respectively.
3.3.2. The idea of implementing LBRQT algorithms ussing cross-layer model
3.3.2.1. Modify the node structure using cross-layer model
Transport
SA
Network
MAC
Physical
Predicting the
parameters of
performance
Update the
database of
traffic density
SA: Stationary Agent
Data
RREQ
Node j
Figure 3.6. Cross-layer model uses for the
LBRQT algorithm
To be able to use information about
QoT for routing constraints, the net-
work layer must be able to directly
access to the information of the
physical layer. This can only be per-
formed by using cross-layer model
[2, 5, 26]. In LBRQT algorithm,
the cross-layer model is proposed as
shown in Figure 3.6, where an sta-
tionary agent (SA) is used for the
exchange of the information of QoT
between physical and network lay-
ers. The tasks if the SA includes: (i)
updating traffic load for the connections in the network, and (ii) predicting the per-
formance parameters which include the blocking probability of the data packets,
SNR of a route and EED. The information of QoT and EED are used for routing
constraints according to (3.21) and (3.22). The information of BPD is used for the
11
criteria of selecting the load balancing route according to the objective function
(3.19) by source node.
3.3.2.2. Improve the processing RREQ and RREP at each node
(i) RC of the intermediate node does not have a valid route to destination
I
S
.
K
L
M
.
.
.
P
RREQ
SA at node I predicts QoT, EED and
BPD from S to each neighbor of node I
RREQ
.
.
.Data Packet
SA at I statistics the load traffic offering
to link from I to the next node
RREQ
The set of all neighbors of node I
The set of all neighbors of node I satisfies the
constraint conditions of QoT and EED (Set Qi)
I
S
.
D
.
M
L
RREQ
RREP
QoT and EED from S to D
don’t satisfy the given
constraint conditions
QoT and EED from S to D
satisfy the given constraint
conditions
RREQ
RREQ
RREQ
SA at I predicts QoT, EED and BPD from S to
D along the route S I joins I D
(a)
(b)
Figure 3.7. Principle of process RREQ when RC
of node I has no route to the destination
This idea is illustrated as Fig. 3.7.
When node I receives an RREQ
packet of route discovery request
from S to D, SA at I predict the
measurements of QoT and EED
from S to each neighbor of I. Then,
SA determines the setQi is a set of
neighboring nodes of I that satisfy
the QoT constraints. Thence node I
only broadcast RREQ to the nodes
of set Qi. In addition, after deter-
mining setQi, SA at I also predicts
BPD from S to each node of set Qi. This BPD is used for source node to select a
load balancing route. The set Qi is determined by Algorithm 3.1.
Algorithm 3.1: Finding set of neighbors of I satisfying constraints of QoT (Set Qi)
(1) Read the information of (β (r)si and τ
(r)
si ) in RREQ;
(2) Qi← /0 ;
(3) for ((each node J is the neighbor of node I) do
(4) Collect the information SNR from I to J (β (h)i j ) at physical layer;
(5) Predict EED from I to J (τ(h)i j ) according to (3.9);
(6) τ(r)s j ← τ(r)si + τ(h)i j ;
(7) if ((Relay type of the nodes is DF) then
(8) β (r)s j ← min(β (r)si ,β (h)i j );
(9) else
(10) β (r)s j ←
(
1/β (r)si +1/β
(h)
i j
)−1
;
(11) end
(12) if ((τ(h)s j ≤ τth) and (β (h)s j ≥ βreq)) then
(13) Read information BPD from S to I (B(r)si ) in RREQ;
(14) Predict BPD of hop from I to J (B(h)i j ) according to (3.7);
(15) B(r)s j = 1− (1−B(r)si )(1−B(h)i j ); Qi← Qi ∪ J;
(16) end
(17) end
12
(ii) RC of the intermediate node has a valid route to destination
I
S
.
K
L
M
.
.
.
P
RREQ
SA at node I predicts QoT, EED and
BPD from S to each neighbor of node I
RREQ
.
.
.Data Packet
SA at I statistics the load traffic offering
to link from I to the next node
RREQ
The set of all neighbors of node I
The set of all neighbors of node I satisfies the
constraint conditions of QoT and EED (Set Qi)
I
S
.
D
.
M
L
RREQ
RREP
QoT and EED from S to D
don’t satisfy the given
constraint conditions
QoT and EED from S to D
satisfy the given constraint
conditions
RREQ
RREQ
RREQ
SA at I predicts QoT, EED and BPD from S to
D along the route S I joins I D
(a)
(b)
Figure 3.8. Principle of process RREQ when
RC of node I has a route to the destination
Figure 3.8 illustrates the idea of im-
proving RREQ processing at each node
when the intermediate node’s RC has
a valid route to the destination node.
Assuming the current node is I, in this
case, node I does not immediately cre-
ate RREP and reply to S as the on-
demand routing protocol. Instead, the
SA at I predict QoT and EED from S to
D along the route S→ I join with I→
D. If predicted QoT and EED satisfy the given constraints, RREP is created and
reply to source node. In contrast, node I proposes RREQ as case (i).
Algorithm 3.2: Predict QoT and BPD by SA when RC of I has a route to D.
(1) Read information of QoT and EED from S to I (β (r)si and τ
(r)
si ) in RREQ;
(2) Read information of QoT and EED from I to D (β (r)id and τ
(r)
id ) in RC of I;
(3) τ(r)sd ← τ(r)si + τ(r)id ;
(4) if (Relay type of the nodes is DF) then
(5) β (r)sd ← min(β (r)si ,β (r)id );
(6) else
(7) β (r)sd ←
(
1/β (r)si +1/β
(r)
id
)−1
;
(8) end
(9) if ((τ(h)s j ≤ τth) and (β (h)s j ≥ βreq)) then
(10) Read information of BPD from S to I (B(r)si ) tin RREQ;
(11) Read information of BPD from I to D (B(r)id ) in RC of I;
(12) B(r)sd = 1− (1−B(r)si )(1−B(r)id ); Create RREP, store B(r)sd into RREP;
(13) else
(14) Find set Qi according to Algorithm 3.1;
(15) end
3.3.2.3. Improve the route selection mechanism at the source node
For the improved process of RREQ and RREP as Section 3.3.2.2, if a route is
found, this route always satisfies the constraints of QoT. The remaining problem
of the LBRQT algorithm is to choose a load balancing route. This is done at the
source node. According to the principle of the LBRQT algorithm, the criterion for
selecting a route is to minimize BPD according to the objective function (3.19).
Therefore, when the RREP packet is received, the source node selecting the route
with the minimum BPD value.
13
3.4. The operation principle of LBRQT algorithm
Start
Discard
RREQ
I is destination (D)
Yes
Yes
No
Yes
No
Determine Qi
according to
Algorithm 3.1
Node I broadcast
RREQ to all node
J Qi
S creates
RREQ
Sai
Yes
Predict QoT and
BPD according to
Algorithm 3.2
Send
RREQ to S
For each J Qi
I = J
I = S
Determine Qi
according to
Algorithm 3.1
Node I broadcast
RREQ to all node
J Qi D create
RREP
Send
RREP to S
NRREP = 0;
Twait = 0;
Increase Twait
S receives RREP
NRREP = NRREP + 1
S selects route with
minimum BPD
Reject request because the
route could not be found
End
Yes
Yes
No
No
No
Yes
No
Yes
No
Source node Intermediate node Destination
node
RC of I has a
route to D?
(NRREP = K) OR
(Twait > Timeout)
NRREP > 0
RREP is
created?
Qi
Qi
I not yet received
this RREQ?
Figure 3.9. Flowchart of LBRQT routing algorithm
3.5. Apply for AODV protocol
3.5.1. Introduction
The research results in Chapter 2 have shown that, for the discovery principle of
AODV, there are some cases where the route found does not satisfy the QoT con-
straint. To solve this problem, the author applied the LBRQT algorithm to improve
the route discovery mechanism of the AODV protocol [16], in order to find the load
balancing route, while satisfying the QoT constraints. The improved algorithm is
named LBRQT-AODV. This proposal of the author has been published in [B2]1.
3.5.2. Modify the format of RREQ and RREP packets
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Type J R G D U Reversed CF Hop Count
(10) RREQ ID
(11) Destination IP Address
(12) Destination Sequence Number
(13) Source IP Address
(14) Source Sequence Number
(15) BP (16) QoT (17) EED
32 bits
(1) (2) (3) (4) (5) (6)
Type J R Reversed Prefix Hop Count
(7) Destination IP Address
(8) Destination Sequence Number
(9) Originator IP Address
(10) Lifetime
(11) BP
(a)
(b)
32 bits
Reversed Reversed
Figure 3.11. Format of (a) RREQ and (b) RREP packets in LBRQT-AODV
1Journal of Communications, Vol.13, No.7, 2018, pp. 338-349 (SCOPUS).
14
3.5.3. LBRQT-AODV algorithm
Algorithm 3.3: LBRQT-AODV algorithm at source node
(1) S creates RREQ;
(2) SA determines Qs according to 3.1;
(3) if (Qi 6= /0) then
(4) Broadcast RREQ to all nodes in Qs;
(5) Wait until receives K of RREP packets or over timeout;
(6) if (Number of received RREP packets > 0) then
(7) Select the route with BPD value in RREP is the smallest RREP to update
into the RC of S;
(8) else
(9) Reject route discovery request;
(10) end
(11) else
(12) Reject route discovery request;
(13) end
Algorithm 3.4: LBRQT-AODV algorithm at intermediate or destination nodes
(1) Node I receives RREQ;
(2) if (I is intermediate node) then
(3) if (I haven’t received this RREQ package before) then
(4) Update the reverse route to S into the RC of I;
(5) if ((RC of I don’t have a valid route to D) then
(6) SA determines Qi according to Algorithm 3.1;
(7) if (Qi 6= /0) then
(8) Broadcast RREQ to all nodes in Qs;
(9) else
(10) Discard RREQ and End the processing RREQ;
(11) end
(12) else
(13) if (DSN of route I→ D is greater than DSN in RREQ) then
(14) SA predicts QoT, EED and BPD along route S→ I join I→ D
according to 3.2;
(15) if (RREP is created) then
(16) Send RREP to S according to the reverse route;
(17) else
(18) Run the steps from 6 to 11;
(19) end
(20) else
(21) Run the steps from 6 đến 11;
(22) end
(23) end
(24) else
(25) Discard RREQ and End the processing RREQ;
(26) end
(27) else
(28) Update the reverse route to S into the RC of I;
(29) Create RREP, send RREP to S according to the reverse route;
(30) end
15
3.6. Apply for DSR protocol
3.6.1. Introduction
The research results in Chapter 2, for the discovery principle of DSR, there are
some cases where the route found does not satisfy the QoT constraint. To solve this
problem, the author applied the LBRQT algorithm to improve the route discovery
mechanism of DSR protocol. The improved algorithm is named LBRQT-DSR.
3.6.2. Modify the format of RREQ and RREP packets
The RREQ and RREP of the LBRQT-DSR are modified as shown in Figure 3.12.
3.6.3. LBRQT-DSR algorithm
Algorithm 3.5: LBRQT-DSR algorithm
(1) S creates RREQ; I← S; NRREP = 0;
(2) repeat
(3) Determine Qi according to Algorithm 3.1;
(4) Broadcast RREQ to all node J in Qi;
(5) if (J has not received this RREQ before) then
(6) Add a record to the RC of J containing the reverse route to S;
(7) if (J is not destination (D)) then
(8) if (RC of J don’t have a route to D) then
(9) Update the reverse route to S into RC of J;
(10) Update the route from S to J into RREQ;
(11) I← J;
(12) else
(13) SA at J predicts QoT, EED and BPD according to S→ I join I→ D
according to Algorithm 3.2;
(14) if (RREP is created) then
(15) Join route S→ J to J→ D;
(16) NRREP ← NRREP+1; Send RREP to S according to reverse route;
(17) else
(18) Update the reverse route to S into RC of J;
(19) Update the route from S to J into RREQ;
(20) I← J;
(21) end
(22) end
(23) else
(24) Create RREP; Update the route S→ D into RREP;
(25) NRREP ← NRREP+1; Send RREP to S according to reverse route;
(26) end
(27) else
(28) Discard RREQ and End the proposing RREQ;
(29) end
(30) until (NRREP = K) or (over timeout);
(31) if (NRREP > 0) then
(32) S selects a route with BPD value in RREP is the smallest;
(33) else
(34) Reject the route discovery request from S to D;
(35) end
16
Opt. type (*) Opt. Data Length (*) Identification (*) Opt. type (*) Opt. Data Len (*) Last Hop Ext. (*) Reserved (*)
Target Address (*) Address [1] (*)
Address [1] (*) Address [2] (*)
Address [2] (*) Address [3] (*)
(*) (*)
Address [n] (*) Address [n] (*)
BP (**) QoT (**) E2E (**) BP (**)
(a)
(b)
Reserved Reserved
Figure 3.12. Format of (a) RREQ and (b) RRREP in LBRQT-DSR algorithm
3.7. Simulate and analyze results
3.7.1. Simulation scenario
LBRQT-AODV and LBRQT-DSR algorithms are evaluated by simulation on OM-
NeT ++ [10], compared to AODV [16], DSR [22] and DSR-SNR algorithms in
[24]. The simulation scenario is set as Section 2.5.1, chapter 2.
3.7.2. Simulation results of LBRQT-AODV algorithm
Figure 3.13. Compare SNR of (a) AODV and (b)
LBRQT-AODV
Figure 3.13 compares SNR
of routes using AODV and
LBRQT-AODV in the case of
the 50 nodes topology, aver-
age mobility speed is 10 m/s.
We can observe that there
are many routes that do not
satisfy the QoT constraints.
For LBRQT-AODV, SNR has
been improved. Most of SNRs
are greater than required SNR (23.5 dB).
0.00
0.01
0.02
0.03
0.04
0.05
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
BP
D
AODV
LBRQT-AODV
Traffic load (Erlang)
Figure 3.17. Compare BPD of
AODV and LBRQT-AODV
As the SNR of LBRQT-AODV algorithm im-
proved, BPD decreased as shown in Figure 3.17.
This result is simulated on the 40 nodes topol-
ogy, the average mobility speed of each node is
5 m/s. When the traffic load is 0.6 Erlang, the
BPD of the AODV is 0.0136. Meanwhile, this
value of LBRQT-AODV is only 0.0091. Thus,
BPD of LBRQT-AODV decreased by 33.21 %
compared to AODV.
62E+6
64E+6
66E+6
68E+6
70E+6
72E+6
74E+6
76E+6
0 50 100 150 200 250 300 350 400 450
AODV
LBRQT-AODV
Simulation time (s)
Th
rou
gh
pu
t (b
it/s
)
Figure 3.18. Compare throughput
of AODV and LBRQT-AODV
For throughput, LBRQT-AODV is also more ef-
ficient than the AODV algorithm. This is clearly
shown in Figure 3.18, corresponding to the case
where the number of nodes is 40, mobility speed
17
5 m/s. The average throughput of the AODV and LBRQT-AODV algorithms are
69.85 and 71.55 Mbit/s, respectively. Thus, compared with the AODV algorithm,
the throughput of the LBRQT-AODV algorithm increases by 1.7 Mbit/s.
3.7.3. Simulation results of LBRQT-AODV algorithm
21.00
22.00
23.00
24.00
25.00
26.00
20 25 30 35 40 45 50
DSR
LBRQT-DSR
Required SNR
Network size (nodes)
M
in
im
um
S
NR
(d
B)
Figure 3.20.Minimum SNR of
LBRQT-DSR and DSR
Figure 3.20 shows the minimum SNR of routes.
For DSR, SNR is greater than required SNR
when the number of nodes is less than 30. How-
ever, if the number of nodes is greater than
30, the SNR is smaller than required SNR. For
LBRQT-DSR, SNR has been improved, always
greater than required SNR despite the number
of nodes is large. For BPD, when using LBRQT-
DSR, BPD is also improved compared to DSR
(Figure 3.23). BPD of LBRQT-DSR decreased
on average 51.79 % compared to DSR.
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
BP
D
DSR
LBRQT-DSR
60E+6
62E+6
64E+6
66E+6
68E+6
70E+6
72E+6
0 50 100 150 200 250 300
LBRQT-DSR
DSR
Traffic load (Erlang) Simulation time (s)
Th
ro
ug
hp
ut
(b
it/s
)
Figure 3.23. Compare BPD of
LBRQT-DSR and DSR
Figure 3.26. Throughput of
LBRQT-DSR and DSR
In terms of through-
put, LBRQT-DSR al-
ways achieves a higher
throughput than the DSR
algorithm (Figure 3.26).
LBRQT-DSR algorithm
yields higher through-
put than the average
DSR by 2.99 Mbit/s.
3.8. Conclusion
Chapter 3 presented the load balancing routing algorithm ensuring quality of trans-
mission (LBRQT), proposed for MANET. LBRQT algorithm finds the route that
satifies the QoT constraints, while balancing the traffic load across all connec-
tions. The LBRQT algorithm has been applied to improve the AODV routing pro-
tocols (LBRQT-AODV) and DSR (LBRQT-DSR). Simulation results on OMNeT
++ showed that the algorithms LBRQT-AODV and LBRQT-DSR have found the
routes that satify the constraints of QoT, so QoT of the data transmission routes is
always guaranteed. In addition, the routes are also selected according to the load
balancing criteria. Therefore, minimi

Các file đính kèm theo tài liệu này:

- improve_the_performance_of_mobile_ad_hoc_network_using_load.pdf