Resilient Packet Transmission (RPT) for Buffer Based Routing (BBR) Protocol

Front Cover

SubmissionArea

(Division)1C (Communication System and Security)

Paper Title Resilient Packet Transmission (RPT) forBuffer Based Routing (BBR) Protocol

List ofAuthors Geetanjali Rathee, Nitin Rakesh

Affiliation2

Geetanjali Rathee: M-Tech ResearchScholar, Nitin Rakesh: Assistant Professor (SG)Department of Computer Science andEngineering

Corresponding AuthorInformation

Name Geetanjali Rathee, Nitin Rakesh

Mobile Phone +91-9736248186, 9218540185

Telephone

Fax

E-mail [email protected], [email protected]

1 Choose one of A, B, C, and D, and write it.2 Write down the affiliation information for all the authors.

1

RPT For BBR Protocol

Address3

Vivekananda Bhawan, Departmentof Computer Science andEngineering, Jaypee Universityof Information and Technology,Waknaghat, Himachal Pradesh,India- 173234

Remark4

Resilient Packet Transmission (RPT)for Buffer Based Routing (BBR)

Protocol

Geetanjali Rathee*, and Nitin Rakesh*

Abstract— To provide effective communication in Wireless MeshNetwork (WMN), several algorithms have been proposed. Since, thepossibilities of numerous failures always exist duringcommunication; resiliency has been proved to be an importantaspect for WMN to recover from these failures. Resiliency ingeneral is the diligence of reliability and availability innetwork. Several types of resiliency based routing algorithmshave been proposed i.e. Resilient Multicast, ROMER etc. ResilientMulticast establishes two-node disjoint path and ROMER usescredit based approach to provide resiliency in the network.However these proposed approaches have some disadvantages interms of network throughput and network congestion. PreviouslyBuffer Based Routing (BBR) approach has been proposed to overcomethese disadvantages. We have proved earlier that BBR is moreefficient w.r.t throughput, network performance and reliability. Inthis paper we have considered the node/link failure issues andanalogous performance of BBR. For this we have proposed Resilient

3 Postal (zip) code should be included.4 If the paper has been recommended from a certain conference/workshop, please provide the conference title and the relevant information.

2 Copyright 2013 KIPSⓒ

Manuscript received August 23, 2013; accepted January 17, 2014. *Dept. of Computer Science Engineering (CSE) & Information Communication

Technology, Jaypee University, Waknaghat, India([email protected], [email protected] )

Geetanjali Rathee, Nitin Rakesh

Packet Transmission (RPT) algorithm as a remedy for BBR duringsuch failures. Further we have shown the comparative performanceanalysis of previous approaches with our proposed approach.Network throughput, network congestion and resiliency againstnode/link failure are particular performance metrics which areexamined over different sized WMN.

Keywords— WMN; Resiliency; BBR; Routing; RPT

1. INTRODUCTION Today, network services (like email, world wide web etc) [1-

2] have become a basic need in day-to-day communication. Forproviding these network services more effectively, WMN (WirelessMesh Network) [3-6] has turned into a popular topology whichbuilds high performance infrastructure. With the growth ofnetwork services, several types of network communication threatsare moreover coming into existence. To resist thesecommunication threats, resiliency [7-13] is a significantapproach for WMN. Basically resiliency is the capability toprovide services in the face of failure. Resilient MulticastRouting [14] and ROMER [15] are the popular resilient routingalgorithms for WMN. Both these approaches are having problems ofrestricted network throughput, network congestion, andsuccessful packet delivery against node/link failure. Let usintroduce both these approaches to classify these problems.Resilient Multicast Routing Protocol [14] establishes two-node

disjoint path to communicate between each[source,destination] pair. In case of node/link failure,traffic on unaffected path reaches the destination. This willincrease network communication cost (reduces throughput).Another drawback of this approach is increased networkcongestion due to failure of a node/link but multiple failureswill further restrict resiliency in the network. While in ROMER [15], source node forwards the packet by taking

maximum credit cost. At each node credit cost R and thresholdvalue T is calculated to reach from source to destination. Themajor drawback of ROMER is that, if cost at each node is higherthen there is possibility to discard the packet. ROMERbroadcasts the packets in the network and this will resultincreased network communication cost (reduces throughput) andincreased network congestion due to multiple packet delivery in

3


the network. In our previous work, we have proposed BBR (Buffer Based

Routing) [16] approach which provides an efficient solution overthe drawbacks of Resilient Multicast and ROMER. In BBR we haveallocated buffers at alternate nodes using BAA (BufferAllocation Algorithm) with the aim to improvise resiliency andto increase the throughput and fast packet delivery in network. This paper is divided in four sections. In first section, we

have introduced the disadvantages of Resilient Multicast andROMER approaches. We have also introduced how BBR approach ismore advantageous over these approaches. In section two, BAAapproach is analysed for buffer allocation and further in thissection calculation of minimum and maximum buffer size duringcommunication from source to destination is also analysed.Furthermore we have proposed RPT algorithm for resilient packettransmission. The issues of time complexities during BAA and RPTare also discussed in this section. Performance evaluation andsimulation results are shown in section third. Finally fourthsection concludes the paper and the future scope of the work isdiscussed.

2. BUFFER ALLOCATION MECHANISM USING BAA2.1 Buffer Allocation Mechanism using Buffer AllocationAlgorithm (BAA)In this paper by using an example we introduce an algorithm

for Resilient Packet Transmission (RPT) which clarifies howpackets are transmitted to destination node and give its timecomplexities. BBR [16] approach adopts routing technique basedon buffer allocation i.e. we provide buffering at each nodeinstead of maintaining routing table. The BBR approach consistsof three steps: i) Buffer allocation to the network nodes usingBAA; ii) Selection of optimum path for routing; iii) ResilientPacket Transmission. The buffer allocation process in thenetwork is discussed as follows.

2.1.1. Buffer Allocation to the network nodes:

According to BBR least cost path selection, buffers are placedat alternate positions in the network. The buffer allocation isachieved in following steps: a) Select a node Ni randomly from network in figure 1,

where i =1, 2, 3……n number of nodes in this network.Assign buffer to this node and mark it as visited node.

4

11

22

111B

C D E

F G

H I J

A

311

22

2

23

4


b) Choose least cost path from node Ni to its connectedneighbouring node and move to this node, make it as Ni.As buffer allocation process is assigned for alternatenodes. So, skip buffer assignment to this node and justmark as visited.

c) Again choose least cost path from node Ni to itsconnected neighbouring node. Move to this node and makeit as Ni. Assign buffer to this node and mark it asvisited node.

d) This step compares whether the next node is visited ornot. If node is visited then it rollbacks to itsprevious node and again search for another node.

e) Repeat steps a−d until total buffer placement in networkis performed. The buffering process will stop when thetotal buffered nodes are ≤ n /2 where n are number of nodesin the network.

Fig. 1. Buffer allocation in the network using BAA (steps¿)).

Buffers are provided to maintain the resiliency in the network.If we have a network of 10 nodes (figure 1) then total number ofbuffers allocated in the network are 5. Similarly in a networkof 15 nodes, number of buffers allocated in network are 7. If(n= number of nodes in the network) Then total buffer placementin the network is ≤ n /2. Minimum and maximum size of buffer isprovided to be 5 units of packet i.e. at a time only 5 packetsare stored inside the buffer. To reduce the traffic congestionand increase the speed of data transfer, the minimum and maximumsize of buffer is taken to be 5 units. To understand it in abetter way let us take an example: figure 1 shows a 10 nodenetwork in which source node A and intermediate node F arebuffered nodes, let there is only one packet to transmit fromsource A to destination node J.

5


Step 1: node A sends first packet to node C, assuming thetransmission of first packet from node A to C is one second.Step 2: node C sends an acknowledgement (ack) of packet 1 tonode A, assuming ack transmission time is also one second.Step 3: further node C transmits the same packet 1 to bufferednode F within same time period (i.e. 1 second).Step 4: buffered node F sends ack to node C (in 1 sec) and thennode C forwards the buffered node F ack to node A (again in 1sec).So, total amount of time to transmit the first packet from nodeA to node F is 5 (1+1+1+1+1) seconds. After 5 seconds firstpacket will be removed from A buffer node and next packet willarrive. To increase the speed of processing and decrease delayand congestion in the network we take buffer size of 5 units. BAA node generation results has been shown in figure 4.Theterminologies used in RPT table 3), BBR (table 4) are shown intable 2.

2.1.2. Selection of Optimum Path for Routing

In BBR approach, the routing table consist of seven parts i.e.1) node; 2) node address; 3) next hop; 4) next hop buffered 5)next hop address; 6) cost; and 7) buffered node. Initiallybuffer space will not be allocated next hop and next hopbuffered field. These fields are updated when buffer allocationalgorithm is executed. When buffer allocation algorithm isterminated, last visited node consists of the updated which itbroadcasts to entire nodes of the network. Thus Routing Table(RT) of each node of the network is updated. Table 1 shows RT ofbuffered node J which is the last step of BAA algorithm beforetermination.

2.1.3. Resilient Packet Transmission (RPT)

Aim of RPT is to successfully transfer information from source to destination node of a network even during failures. For this purpose, routing path from source to destination in RPT must have following characteristics [16]:a) The route must contain minimum number of buffered node.

6


b) If more than one path has same number of buffered nodes then it will select least cost.

Initially BBR approach allocates buffers to the entire network(as described in part (i)) and after that packet transmissionstarts. In the network less than or equal to n /2 nodes arebuffered. During packet transmission, the non buffered nodeforwards packet to next node and send acknowledgement (ACK) toits preceding node. The preceding buffered node will storepacket until ACK is received from next buffered node. When theACK is received from next buffered node, the preceding bufferednode deletes the packet from the buffer. The detailedexplanation of packet transmission and failure cases has beenalready discussed in our previous paper [16]. Table 3 shows theproposed Resilient Packet Transmission (RPT) algorithm.

Table 1. Routing Table of Buffered Node ‘J’Node

IDNode

address

Nexthop ID

Next hopbuffered?

Nexthop

address

Cost Bufferednode?

A … C No … 1 Yes

A … D No … 2 Yes

B … D No … 1 Yes

B … G No … 2 Yes

B … E No … 1 Yes

C … F Yes … 2 No

C … H No … 3 No

D … A Yes … 2 No

D … B Yes … 1 No

D … F Yes … 2 No

D … G No … 1 No

E … B Yes … 1 No

E … J Yes … 2 No

F … C No … 2 Yes

F … D No … 2 Yes

F … H No … 4 Yes

F … I No … 2 Yes

G … D No … 1 No

G … B Yes … 2 No

7


G … I No … 1 No

G … J Yes … 1 No

H … C No … 3 No

H … F Yes … 4 No

I … F Yes … 2 No

I … G No … 1 No

J … G No … 1 Yes

2.2 COMPLEXITY ANALYSIS

This section shows the time complexity of proposed BAA [16] andRPT approach. BBR approach is previously allocating buffer tillthe destination but now, we have updated the BBR approach towork for un-traversed link beyond destination. The timecomplexity of both BAA and RPT is executed asO¿ using BruteForce method. Table 4 and 5 shows the time complexity of BAA andRPT approaches.

Table 2: Terminologies used in BAA and RPT Algorithms

Terms used inalgorithms

(table 3,4 and5)

Description

Buffered node Ni Signifies the node which is a buffered node (i.e. alreadyassigned a buffer)

Next[Ni] It is used to visit the next node in the network duringbuffer allocation process

(Node→Node_next)

During allocation of buffers in the network, we maintaina list of nodes that are visited and assigned buffers.So, Node→Node_next which defines: if next node (i.e. node→node_next) is present in visited list then rollback toits parent node (i.e. previous→Ni )

Buffer[Ni] Signifies the array which contains node id of thebuffered node

visited list[ ] It is an array which contains the node id and stores theinformation of previously visited node in the network

Ni←min_cost_next[Ni]

It is used to select the node which has minimum link costfrom the current node and then the node which has lessminimum cost will become the next current node

Ni !=visited_list[ ]

In our algorithm for each next node Ni we check whether Ni

is present in visited list. If Ni is present in visitedlist array then we go for next node (which has

8


‘next_min_cost(Ni)) otherwise we continue our processfrom current node

Bufferedarray[ ]

It is an array which contains node id of the nodes whichhas already been buffered

visited list[ ]+ 1 = Ni

It maintains visited current node Ni into visited_list[]

assign_buffer(Ni)

It is a function used to assign buffer to the node Ni

rollback(previous→Ni)

If node Ni is present in visited list and its next nodealso present in visited list then we rollback to parentnode of Ni

array_buffer[ ]←Ni

It is an array which contains the information about thebuffered node, if the node Ni is found eligible to assignthen it is added in this array

update_routing_table()

It is a function which is used to update the routingtable

Table 3. Resilient Packet Transmission Algorithm

1. Source (buffered) node starts transmission through least cost path byconsidering its routing table ( in accordance to section 2.1.3 (a) and(b))

2. Wait for two ack’s (buffered node/non buffered node)

3. If (next[Ni ]!=buffered node Ni ) 4. Receive and forward the packets to its downstream node Ni and

starts RTT (round trip time).5. Send ack to its previous node Ni.

6. Go To step(3) 7. End: if8. Else If (next[Ni]=buffered [Ni]) 9. Store and forward the packet to next N i and

starts RTT 10. Send ack to its previous node Ni

11. Wait for two ack’s. 12. If (two ack’s received within RTT) 13. Go To step(8)14. Else15. Go To step(22) 16. End: else if17. Else18. If ( Packet receive within its RTT) 19. Repeat step 3

9


20. End: if21. Else22. Failure (Ni) 23. End: else24. End: elseif25. Else (destination node) 26. Receive and store packet 27. Send ack to its previous node. 28. End: else

Failure (Ni) 1. If (failure Ni=exist) 2. Preceding buffered node select another least cost path to

destination node.

3. End if

Table 4. Time Complexity of Buffer Allocation Algorithm

1. Select random node Ni from the network O(1)2. Assign buffer (Ni) O(1)3. Visited list [] =Ni O(log2 (N ))4. If (next [Ni]! =null) O(1)5. Ni←min_cost_next [Ni] O(N)6. If (Ni=visited list []) O(log2 (N ))7. Select next_min_cost (Ni) O(1)8. Go To step (4) O(Nlog2 (N ))9. End if10. Else

11. If (previous [Ni] =buffered array []) O(log2 (N ))12. Skip (Ni) O(1)13. Visited list [] +1=Ni O(log2 (N ))14. Go To step (4) O(log2 (N ))15. End if16. Else

17. Allocation(Ni) O(Nlog2 (N ))18. Goto step(4) O(Nlog2 (N ))19. End else

10


20. End else21. Else22. Rollback (Ni); O(1)23. End else24. Repeat step 2 until two rollbacks occur

at the same node;O(1)×Nlog2 (N )

25. Return;Allocation (Ni) Begin

1. If (Ni!=visited_list[]) Olog2 (N)2. assign_buffer(Ni) Olog2 (N)3. visited_list [] +1=Ni O(1)4. End if

5. Else6. rollback (previous→Ni) O(1)7. End else8. End

assign_buffer (Ni) Begin

1. assign buffer to Ni O(1)2. array_buffer []←Ni O(1)3. update_routing_table() Olog2 (N)4. End

Total complexity of Buffer allocationalgorithm:

O¿

Table 5. Complexity analysis of RPT

1. Select random node Ni from the new Source (buffered) node starts transmission through least cost path by considering its routing table

O(log2 (N ))

2. Wait for two ack’s (buffered node/non buffered node) O(1)

3. If (next[Ni ]!=buffered node Ni O(1)4. Receive and forward the packets to

its downstream node Ni and starts RTT(round trip time)

O(1)

5. Send ack to its previous node Ni O(1)

11


6. GoTo step(3) O(N)7. End: if8. Else If (next[Ni]=buffered [Ni]) O(1)9. Store and forward the packet to next

Ni and starts RTT O(log (N))

10. Send ack to its previous node Ni O(1)11. Wait for two ack’s O(1)12. If (packet received within RTT)13. Go To step(8) O(N)14. Else15. Go To step(22) O(N)16. End: elseif17. Else18. If (Packet receive within its RTT) O(1)19. Repeat step 3 O(N)20. End: if21. Else22. Failure(Ni)23. End: else24. End: else if

O(log (N))

25. Else (destination node)26. Receive and store packet27. Send ack to its previous node28. End: else

O(1)O(1)O(1)

Failure (Ni)1. If (failure Ni=exist) O(1)2. Preceding buffered node select

another least cost path to destination node.

3. End if

Olog2 (N)

Total complexity of RPT algorithm: O¿

The time complexities of Buffer Allocation and RPT algorithmsareO¿.

3. PERFORMANCE EVALUATION

12


In this section we have analyzed the performance of BBRapproach with previously defined approaches i.e. resilientmulticast [14] and ROMER [15]. We evaluate network performanceusing some parameters i.e. throughput (defined in terms ofcost), network congestion (in terms of packet transmission) andresiliency against node/link failure (in terms of faulttolerance). We have considered cost as total amount of delayoccurs during transmission of packets from source todestination. While packet transmission is the total number ofpackets transmitted at a time in network and fault tolerance[17-19] is possible numbers of paths exist after failure. Now,evaluating the performance of Resilient Multicast, ROMER and BBRover five different network sizes i.e. 5,10,15,20,25,100.Let us consider first, a network of five nodes (see Fig. 2(a))where A is the source node, E is the destination node andcompare between these approaches.a. Network Throughput: Throughput is defined in terms of cost

i.e. total amount of delay occurs during transmission ofpackets from source to destination. Let us evaluate networkthroughput of these approaches for network size 5 (Fig.2(a)).

i. Network throughput of Resilient Multicast [14]: Thisapproach selects at least two node disjoint paths to senddata packets. To analyze throughput, we calculate thetotal amount of cost of selected disjoint paths to reachfromS−D (see Fig. 2(a)). SourceA selects two disjointpaths i.e. i)A−B−D−E which consumes 1+2+2=5 units(throughA−B,B−D,D−E) and ii)A−C−E which consumes2+1=3 unit of cost (through A−C, C−E). So, packet issent to its destination node by consuming (cost of path 1+ cost of path 2) 5+3=8 units.

ii. Network Throughput of ROMER [15]: This approach forwardsthe packet by taking maximum credit cost (which isassumed at the source node). ROMER states that each nodehas some cost and packets will be forwarded by every nodeafter calculating the value of credit cost R andthreshold value T. Such that : If (R>T) Then node forwards the packet; Else Discard the packet;The detail steps of this approach for network size 5are

as:a)Initially ‘A’ broadcasts data to its downstream

nodes i.e. B andC, which consume total3 unit of cost

13


to send the packets (A−C = 2 unit) and (A−B = 1unit).

b)Assuming cost of node B=55 and nodeC=50. Packetwill be forwarded by downstream nodes aftercalculating the value of R andT. At node B if R<Tthen it discards the packet. At nodeC, assume R>Tthen it forwards the packet to its downstream nodes DandE. So the total cost of sending the packets fromC−Dand C−E is 1+1=2 units.

c)Further value of R and T are calculated at node Dthen it forwards the packet to destination E afterconsuming 2unit of cost.

d)Now cost to send the packets from A−E is (cost in(i) + cost in (ii) + cost in (iii)) i.e.; 3+2+2=7units of cost.

iii. Network Throughput of BBR [16]: According to our proposedapproach packets are sent through a route which has leastnumber of buffered nodes. In Fig. 2(a), the path A−C−Econtains minimum buffered nodes i.e. 2. So it sends thepacket using A−C−E path by consuming 2+1=3 units(cost of A−C +C−E). Hence, BBR approach takes lessamount of cost to send data packets fromS−D.

b. Network Congestion: It is defined in terms of rate ofpacket transmission i.e. total number of packetstransmitted at a unit of time in a network. Let us evaluatenetwork congestion on these approaches.

i. Network Congestion on Resilient Multicast: Fig. 2(a)shows that sourceA has20 packets to transmit todestination node. [14] Sends the packets through two-nodedisjoint paths i.e.A−B−D−E andA−C−E. Twenty packetswill be sent through either of these path i.e. A−C−E.Resilient Multicast uses redundant copy of these packetsto send through other path i.e. A−B−D−E. If sourcenode has20packets to transmit, [14] will send 40 packetsin network.

ii. Network Congestion on ROMER [15]: To provide successfuldelivery of packets to the destination node. It deliversredundant copy of packets in the network.(see Fig. 2(a))let source node A has 20 packets to transmit. Tocalculate the cost consider following steps:a) ‘A’ forwards traffic to both nodes i.e. B and C,

so total number of packets are 20+20=40.b) ‘B’ discards the packet in case(R<T) and node C

forwards redundant copy of packets to DandE. Now,

14


numbers of packets to transmit are 20+20=40packets.

c) ‘D’ forwards the packets through single path i.e.D−E which transmits 20 packets.

d) Total number of packets inside the network is40+40+20=100 packets (traffic size of ((i) +(ii) + (iii)).

iii. Network congestion on BBR [16]: It forwards the packetsaccording to its buffer capacity as shown in Fig. 2(a).Buffers are placed at alternate positions. Each bufferhas size of 4units. If Source node ‘A’ has 20 units totransmit then only 8 packets will be forwarded inside thenetwork at a time.

c. Network Resiliency: Resiliency against node/link failure ismeasured in terms of fault tolerance. Fault tolerance isdefined as possible number of paths exists after failure.Let node B has failed.

i. Resiliency in Resilient Multicast [14]: Resiliencyachieved in [14] through two-node disjoint path. (SeeFig. 2(a)) source‘A’ selects two pathsA−B−D−E andA−C−E. If node‘B’ fails, then resiliency is achievedby using second path i.e.A−C−E. If nodeB andC fails,then there is no possible way to reach to destination.

ii. Resiliency in ROMER [15]: Resiliency achieved in [15] byforwarding the redundant copy of packets in the network.Initially [15] forwards the packets through threepossible path i.e.A−B−D−E,A−C−D−E,A−C−E. If nodeBfails then there exist 2 possible paths i.e. A−C−D−EandA−C−E.

iii. Resiliency in BBR [16]: Resiliency achieved in BBR bystoring data packets in buffers placed at alternatepositions in the network, when failure occurs in thenetwork during transmission of data packets, thepreviously buffered node selects another efficient pathfor packet transmission (see Fig. 2(a)). If node ‘B’fails then there is no effect inside the network becausewe send the data packets through A−C−E path.

To evaluate the accuracy of performance evaluation ofResilient Multicast, ROMER and BBR we have analyzed thenetwork parameters (throughput, network congestion,resiliency) on 10,15,20,25 network sizes (see table 6).

15

12

2

11

1

2

1D

B C E

G

J KI

A

31

2

1

2

1

2

1

1

L N

M O

31

F H

1

1

2 1 1

55 50

50

2B

B C

D

A1

2 1

2 1

E11

22

111B

C D E

F G

H I J

A

3

23

2

23

4

1

11B

C D E

F

A

12

2

G

X

2

H I J K2 1

L M N O1 2 P

12 R S T WQ

U V

Y

1 1 122

2

1 11 1 1

1 2 2

1 11 1

11

1

1

2

2

13

2

1

1B

C D E

F

H I

A

31

1

2

2

11

LO

T

31

J

G

21

K

M

2

1N P R

2

S2 Q

1

112 2

2

1


Table 6. Parameter growth of three approaches in a network of fivenodes.

ParameterMetric

ResilientMulticast

ROMER BBR

Throughput 8 7 3Network

congestion40 100 8

Resiliency 1 2 1

Fig. 2. Network with different sizes ((a), (b), (c), (d) and (e)represents 5,10,15,20 and 25 network sizes).

16

(a)

(b)

(c)

(d)

(e)


For network size 10: we analyzed network parameters(throughput, network congestion and resiliency against node/linkfailure) on three approaches i.e. Resilient Multicast [14],ROMER [15] and BBR [16] using a network of 10 nodes.See Fig. 2(b), As throughput measures in terms of cost. [14]

selects two-disjoint paths i.e. A−C−F−I−G−J which consumes7 units of cost and A−D−B−E−Jwhich consumes 6units of cost.So, the total amount of cost consumed by Resilient Multicast is13 unit. In [15] (let node C,F,G discards the packet) so, itconsumes12 units of cost(A−C, A−D, D−F, D−B, B−G, B−E,E−J) and send packets through A−D−B−E−J path while BBR[16] takes 7 units of cost through A−D−B−E−J path. Networkcongestion is measured in terms of packet transmission. Supposesource node A has 20 packets to transmit, [14] transmits 40number of packets using A−C−F−I−G−J and A−D−B−E−Jpaths inside the network, [15] transmits 140 packets throughA−D−B−E−J path (assume node C,F,G discards the packet)while BBR [16] has only 4packets to transmit throughA−D−B−E−J path .While resiliency is measured in terms ofpossible path exist after failure. In [14], if node D fails thenpacket will be forwarded throughA−C−F−I−G−J. In [15] thereexist 2 possible paths i.e. A−C−H−F−I−G−J andA−C−F−I−G−J. In BBR, packets will be forwarded throughA−C−F−I−G−J (Explanation has been described in5nodes ofnetwork). Table 7 shows the parameter growth of three approachesin a network of ten nodes.

Table 7. Parameter growth of three approaches in a network of tennodes

ParameterMetric

ResilientMulticast

ROMER BBR

Throughput 13 12 7Network

congestion40 140 16

Resiliency 1 2 1

For network size 15 (see Fig. 2(c)) throughput of ResilientMulticast, ROMER and BBR is 15, 14 and 6, network congestion is40, 180 and 20 and resilient paths are 1, 8, 1. For networksize 20 (see Fig. 2(d)) throughput is 19, 16, 9, networkcongestion is 40, 200, 24 and resilient paths are 1, 5, 1. Fornetwork size 25 (see Fig. 2(e)) throughput is 20, 20, 7,network congestion is 40, 300, 28 and resilient paths are 1, 8, 1 as shown in table 8. Fig. 3(a) shows the comparative

17


analysis of network throughput for Resilient Multicast, ROMERand BBR approach, while Fig. 3(b) shows the robustness of BBRapproach during network congestion. Finally we have seensimulation of 10,100 node network

(a) (b)Fig. 3. Comparative Analysis of Network Throughput and Congestion inResilient Multicast, ROMER and BBR for different network sizes.

(a) (b)

18

5 Nodes 10 Nodes 15 Nodes 20 Nodes 25 Nodes---01234567891011121314151617181920

Size of Network (in term s of Nodes)

Throughput (in terms of Cost)

Resilient M ulticast (RM )ROM ERBuffer Based Routing (BBR)

5 nodes 10 nodes 15 nodes 20 nodes 25 nodes0

50

100

150

200

250

300

Size of Network (in term s of Nodes)

Network Congestion

(in terms of Packet transmission)

BBRResilent M ulticastRO M ER


(c) (d)

Fig. 4. Simulation result of BAA for (a) n=5, (b) n=10, (c) n=25, and(d) n=100 network sizes.

Table 8. Network parameters comparison on three approaches ([14],[15], [16])

Parameter Metric

Approaches 5-node

10-node

15-node

20-node

25-node

100-node

Throughput (in terms of cost)

Resilient Multicast 8 13 15 19 20 89

ROMER 7 12 14 16 20 76BBR 3 7 6 9 7 66

Network Congestion (in terms of Packet transmission)

Resilient Multicast 40 40 40 40 40 40

ROMER 100 140 180 200 300 650

BBR 8 16 20 24 28 88

Resiliency (against node/link failure)

Resilient Multicast 1 1 1 1 1

1

ROMER 2 2 8 5 8 38BBR 1 1 1 1 1 1

4. CONCLUSION AND FUTURE WORKIn this work we have considered node/link failure issues to

provide analogous performance of BBR approach. We have proposedRPT algorithm as an antidote during such failures. We haveevaluated the time complexity of RPT approach as O¿. Further wehave proved that performance of BBR is comparatively improvedover previously proposed approaches i.e. Resilient Multicasting

19


and ROMER using network throughput, resiliency against node/linkfailure, and network congestion as parameters for5,10,15,20,25 nodes network and show the simulation resultsof 10 and 100 node network (in fig 4). In future to achievemore accurate results of BBR approach, we will apply it inphysical environment and will study the results for any otherpossible measures.

REFERENCE

[1]. Kwon Ohbyung, and Yixing Wen, "An empirical study of the factorsaffecting social network service use," Elsevier Journal of Computers inHuman Behavior, vol. 26, no. 2, 2010, pp. 254–263.

[2]. Hu Tao, Robin S. Poston, and William J. Kettinger, "Nonadopters ofOnline Social Network Services: Is It Easy to Have Fun Yet ?",Journal of Communications of the Association for Information Systems, vol. 29, no. 1,2011, pp. 441–458.

[3]. Manoufali, Mohamed, et al, "Technologies and networks supporting maritimewireless mesh communications", Proceedings of the IEEE 6th joint Conference on IFIPWireless and Mobile Networking Conference (WMNC), Dubai, 2013, pp. 1–8.

[4]. Won-Suk, kim and Sang-Hwa Chung, "Design of Optimized AODV Routing Protocolfor Multi-Interface Multi-Channel Wireless Mesh Networks", Proceedings of the IEEE 27thInternational Conference on Advanced Information Networking and Applications (AINA),Barcelona, 2013, pp. 325–332.

[5]. L. Song and Z. Bing Xia, "An Anycast Routing Protocol for Wireless Mesh AccessNetwork", Proceeding of the IEEE WASE International Conference on InformationEngineering (ICIE) , Taiyuan, Shanxi, vol. 2, 2009, pp. 82–85.

[6]. Gupta, Bhupendra Kumar, B. M. Acharya, and Manoj Kumar Mishra,"Optimization of routing algorithm in wireless mesh networks",Proceedings of the IEEE World Congress on Nature & Biologically Inspired Computing(NaBIC), Coimbatore, 2009, pp. 1150–1155.

[7]. Jing Dong, Reza Curtmola, Cristina Nita-Rotaru, "Secure High-Throughput Multicast Routing in Wireless Mesh Networks", Proceedingsof the IEEE Transactions on Mobile Computing, vol. 10, no. 5, 2011, pp. 653–668.

[8]. Helen Herrman, Donna E Stewart, Natalia Diaz-Granados, Elena LBerger, Beth Jackson, Tracy Yuen, "What is resilience?", CanadianJournal of Psychiatry, Revue canadienne de psychiatrie, vol.56, no.5, 2011, pp.258–265.

[9]. Hossen Mustafa,Xin Zhang,Zhenhua Liu,Wenyuan Xu,Adrian Perrig,"Jamming-Resilient Multipath Routing", Proceedings of the IEEETransactionson on Dependable and Secure Computing vol. 9, no. 6, 2012,pp.

20


852–864[10]. Nitin Rakesh and Vipin Tyagi, “Failure Recovery in XOR’ed Networks”,

Proceeding of the 2012 IEEE International Conference Signal Processing, Computing andControl (ISPCC), Waknaghat, India, 2012, pp. 1–6.

[11]. Nitin Rakesh, and Vipin Tyagi. "Failure Detection using Contour Approachon Network Coded Parallel Networks", Proceeding of the International Conference onModelling Optimization and Computing (ICMOC-2012), Elsevier ProcediaEngineering, Kanyakumari, India, vol. 38, 2012, pp. 763–770.

[12]. Seungjoon Lee, Bobby Bhattacharjee, Aravind Srinivasan, SamirKhuller. "Efficient and Resilient Backbones for Multihop Wireless Networks",Proceeding of the 2008 IEEE Transactions on Mobile Computing, vol.7, no.11, pp.1349–1362.

[13]. Bu, Tian, Mun Choon Chan, and Ramachandran Ramjee,"Connectivity, performance, and resiliency of IP-based CDMA radioaccess networks", Proceeding of the 2006 IEEE Transactions on MobileComputing, vol. 5, no. 8, pp. 1103–1118.

[14]. Xin Zhao , Jun Guo , Chun Tung Chou , Sanjay K. Jha "Resilientmulticasting in wireless mesh networks", Proceeding on 13th International Conference onTelecommunications, Polo de Aveiro, Portugal, 2006, pp. 1-4.

[15]. Yuan Yuan, Hao Yang, Starsky H. Y. Wong, Songwu Lu, WilliamArbaugh "ROMER: Resilient Opportunistic Mesh Routing for Wireless Mesh Networks",Proceeding of the first IEEE Workshop on Wireless Mesh Networks (WiMesh),vol. 12,2005.

[16]. Geetanjali Rathee, Ankit Mundra, Nitin Rakesh, S. P. Ghrera“Buffered Based Routing Approach for WMN”, Procceding in IEEE InternationalConference of Human Computer Interaction, Chennai, India, 2013 (accepted inpress).

[17]. Bhagvan Krishna Gupta, Ankit Mundra, and Nitin Rakesh, "FailureDetection and Recovery in Hierarchical Network Using FTNApproach", Proceeding of the International Journal of Computer Science, vol. 10,no. 2,2013, pp. 414-421.

[18]. Kinjal Shah, Gagan Dua, Dharmendar Sharma, Priyanka Mishra andNitin Rakesh “Transmission of Successful Route Error Message(RERR)in Routing Aware Multiple Description Video Coding Over MobileAdhoc Network”, Procceding of the International Journal of Multimedia & ItsApplications, vol. 3, no. 3, 2011, pp. 51–59.

[19]. Nitin Rakesh, and Vipin Tyagi, "Parallel Architecture Coding:Link Failure–Recovery Mechanism (PAC: LF–RM)", Proccedings of theInternational Journal of System Assurance Engineering and Management”, vol. 38,2012, pp.1–11.

21


Geetanjali RatheeShe is a Teaching Assistant (TA) in the Department ofComputer Science Engineering (CSE) & InformationCommunication Technology (ICT) with Jaypee University ofInformation Technology (JUIT), Waknaghat,Solan-173234,Himachal Pradesh,India. She received her B.TechDegree in Computer Science and Engineering (CSE) from

Bhagwan Mahavir Institute of Engineering and Technology (BMIET), Haryanain the year 2011. Now she is undertaking the Master Degree course undersupervision of Dr. Nitin Rakesh in Jaypee University of Information andTechnology (JUIT), Waknaghat, Solan-173234. Her research interestsinclude Resiliency in Wireless Mesh Networking, Routing Protocols, andNetworking.

Nitin RakeshHe is an Assistant Professor (Senior Grade) in theDepartment of Computer Science Engineering &Information Communication Technology with JaypeeUniversity of Information Technology (JUIT), Waknaghat,Solan–173234, Himachal Pradesh, India. He received his

22


Doctorate in Department of Computer Science and Engineering from JUIT,Waknaghat in 2012. In 2007, he received his Master of Technology Degreein Computer Science and Engineering from Jaypee Institute of InformationTechnology, Noida, India and received Bachelor in Technology Degree inInformation Technology from AEC, Agra in the year 2004. He is a memberof Institute of Electrical and Electronics Engineers (IEEE-USA), LifeMember Computer Society of India (LMCSI), and International Associationof Engineers (IAENG). His research outlines emphasis on Network Coding,Interconnection Networks & Architecture, Fault–tolerance & Reliability,Networks–on Chip, Systems–on–Chip, Network Algorithms, ParallelAlgorithms and Fraud Detection, Online Phantom Transactions.

23

T

3

Documents

Resilient Packet Transmission (RPT) for Buffer Based Routing (BBR) Protocol