Improve the performance of mobile ad hoc network using load balancing routing technology ensuring quality of transmission

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