23
CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICE

CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

  • Upload
    others

  • View
    5

  • Download
    0

Embed Size (px)

Citation preview

Page 1: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

CHAPTER 5

MINGLING MULTIPATH WITH QUALITY OF SERVICE

Page 2: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

MINGLING MULTIPATH WITH QUALITY OF SERVICE

5.1 INTRODUCTION

11 As discussed in the previous

chapter, considering the benefits of multipath and requirement of QoS, QoS can be

successfully implemented with multipath architecture. Multipath routing is capable of

aggregating the resources of multiple paths and reducing the blocking capabilities in QoS

oriented networks, allowing data transfer rate at higher rate, when compared to a single path.

It also increases the reliability of delivery. The benefits of provisioning multiple QoS paths

are reliable QoS support and uniformly balanced network load.

This chapter investigates the ways of implementing multipath strategy in QoS domain,

discussed in literature. The approaches considered are based on Dijkstra algorithm, Bellman-

Ford algorithm, Resource reservation & MPLS. In addition to it, the chapter proposes a

multipath QoS approach based on Dijkstra algorithm which is an extension to the single path

algorithm presented in [93]. The proposed algorithm is implemented in MATLAB and

appended as Appendix-A.

5.2 MULTI PATH in QoS DOMAIN

On the basis of literature survey, the areas that have implemented QoS and multipath

collectively are- Dijkstra algorithm, Bellman-Ford algorithm, Resource reservation & MPLS.

5.2.1Using Bellman-Ford algorithm

In considering with multiple constraints, it has been noticed that the Bellman-Ford (BF)

algorithm can potentially solve a two metric routing problem [38]. It is a property of the BF

algorithm that, at its hth iteration, it identifies the optimal path between the source and each

destination, among paths of at most h hops. It searches for a minimum path cost (or maximal

bandwidth) in ascending order of number of hops. The cost of a path is a function of its

available bandwidth i.e. the smallest available bandwidth on all links of the path, and

finding a minimum cost path amounts to finding a maximum bandwidth path. However,

Page 3: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

because the BF algorithm progresses by increasing hop count, it inherently provides for the

hop count of a path as a second optimization criteria. Therefore, the result of the algorithm

comes with the path which is the one with maximal available bandwidth among all the

feasible paths with the minimum number of hops.

Thus, Bellman-Ford algorithm is very powerful in solving most multiple constrained routing

problems if the minimum hop is the main objective function. BF algorithm is also capable of

solving delay, delay jitter, loss & bandwidth constrained routing problem [15] .This Bellman-

Ford algorithm is single QoS path computation algorithm. This single path computation

algorithm has been extended to perform multiple QoS computations [57] .Each of the QoS

metric is manipulated in the same way as in single path algorithm by increasing hop count.

The algorithm has been designed in order to improve fault tolerance & load balancing. The

multiple path computation algorithm searches for maximally disjoint paths (minimally

overlapped paths) so that impact of link failure is reduced & links are more evenly utilized by

spreading the network load over multiple paths. In order to search for multiple alternate paths

to provide fault tolerance and load balancing yet satisfying QoS constraints, it has defined

alternate paths with the following conditions –

-Satisfying given QoS constraints

-Maximally disjoint from already computed paths

-Minimizing hop count.

The algorithm searches for multiple maximally disjoint paths (i.e. with the least overlap with

each other) such that the failure of a link in any of the paths will still leave one or more of the

other paths operational with high probability.

Based on this algorithm, [56] has presented the multiple QoS path algorithm with some

enhancements for maximally disjoint paths i.e. PDMA (partially disjoint multiple QoS path

algorithm with multiple iteration) and also presented FDMA (fully disjoint multiple QoS path

algorithm).

5.2.2 Using Dijkstra Algorithm

Dijkstra algorithm can generate a minimum hop path that can accommodate the required

bandwidth and has maximum bandwidth. However, it requires some modification in order to

be able to calculate the minimum hop path computation as it does not search for a minimum

Page 4: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

path cost in ascending order of hops as Bellman-Ford does. The modification required for

supporting them is straightforward. Firstly on a graph from which all edges, whose available

bandwidth is less than that requested bandwidth, are removed. This can be performed either

through a pre-processing step, or while running the algorithm by checking the available

bandwidth value for any edge that is being considered. Another modification to a standard

Dijkstra based minimum hop-count path computation, is that the list of equal cost next

(previous) hops which is maintained as the algorithm proceeds, needs to be sorted according

to available bandwidth. This is to allow selection of the minimum hop path with maximum

available bandwidth. Alternatively, the algorithm could also be modified to, only keep among

equal hop count paths the one with maximum available bandwidth. This would essentially

amount to considering a cost that is function of both hop count and available bandwidth [38].

Whereas, Bellman-Ford algorithm used in computing k shortest paths ignores equal hop

multiple count. Equal hop multiple paths impair the performance of routing protocol if the

link state information is inaccurate.

[35] has proposed an algorithm DIMCRA (link-disjoint multiple constraints routing

algorithm). The objective of DIMCRA is to find link-disjoint multi-constrained paths

between a pair of source and destination nodes. DIMCRA is based on SAMCRA.

[45] has presented a bandwidth constrained multipath routing algorithm based on Dijkstra.

Besides specifying the optimality criteria that defines the best paths, it has also described the

construction and selection strategies for multiple paths. The path construction is performed

each time link state information becomes available, while the path selection is performed for

each connection request .This algorithm is based on [93].

This algorithm is based on k best one to one non-loop less paths. It has modified the

algorithm to find k one to all loop less paths instead of one to one non-loop less paths. The

algorithm generates k paths that are either shortest in terms of number of hops (hop based) or

widest in terms of bottleneck bandwidth (bandwidth based). It has considered two metrics

bandwidth & hop count. It has developed two categories of k-shortest algorithms-hop based

& bandwidth based and five selection algorithms. It has concluded that hop-based algorithm

outperforms the bandwidth-based one. It has also presented five path selection schemes as

Best-K-Widest (BKW),Random-K-Widest (RKW), Shortest-K-widest (SKW), Best-K-

Shortest (BKS), Widest-K-Shortest (WKS).

Page 5: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

5.2.3 Using Resource Reservation

In order to use the network resources efficiently, bandwidth reservations are made to ensure

high probability of data arrival to its destinations.

Bandwidth reservation means that certain amount of bandwidth is reserved for a program.

When the program accesses the network, then up to that much bandwidth is guaranteed to be

available to the program. By reserving bandwidth, it is possible to provide reasonable levels

of QoS. The idea is to identify traffic flows, which are streams of packets (voice, video, file

transfers, and so on) going to the same destination IP address and port number. Reservations

are negotiated with each network device along a route to a destination. If each device has

resources to support the flow, a reserved path is set up.

[16] has presented a QoS routing algorithm in which the QoS constraint is specified by

bandwidth guarantee. Therefore, the goal of QoS routing algorithm is to find the best path

with sufficient bandwidth. Here the best path is the path with least cost. To find the least cost

path, the algorithm can run either Dijkstra or Bellman-Ford shortest path algorithm.

[24] has presented a family of algorithms that route and reserve resources along parallel sub

routes i.e. fast algorithm, slow algorithm and super fast algorithms.

[76] has presented a framework in which bandwidth can be reserved on the communications

links, and, once reserved, is guaranteed for the required time. The algorithms finds multi

paths, consisting of possibly overlapping paths, based on the available bandwidths on various

links of the network. It considers delay as second consideration criteria. It has presented the

algorithms based on two problems. The first problem requires that a message of finite length

be transmitted from s to d within r units of time. The second problem requires that a

sequential message of r units be transmitted at a rate of n such that maximum time difference

between two units received out of order is not more than q.

5.2.4 Algorithms for MPLS

Multi-Protocol Label Switching (MPLS) networks do routing based on connections. QoS

routing and traffic engineering goals are normally achieved by finding optimal or near-

Page 6: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

optimal explicit routing MPLS algorithms. The MPLS can provide fast packet forwarding. It

is a new Internet technology that is rapidly emerging as a core technology for next generation

networks. MPLS uses a technique known as label switching to forward the data through the

network. Multi-protocol label switching (MPLS) has many attractive features. First, the

MPLS can create easily an explicit-route label switched path as needed and it can easily map

traffic trunks that consist of traffic flows with similar characteristics of traffic requirements.

[18] proposed two multipath constrained based routing algorithms using MPLS. The

algorithms proposed find multiple LSPs(Label-Switched Path) between ingress and an egress

LSR(Label Switch Router) satisfying a given bandwidth constraint. When there is no single

path through the network satisfying a whole bandwidth constraint, the suggested algorithms

divide the bandwidth constraint into two or more sub-constraints and find a constrained path

for each sub-constraint. First, the algorithms calculate the least-cost path between source and

destination, and allocate the path bandwidth (minimum bandwidth of all links along the path)

to that path. Next, it calculates the next shortest path through the network after removing

links having no available bandwidth and allocates the path bandwidth to that path. The

process is continued until it can allocate the whole bandwidth constraint to the successive

shortest paths. If a single path satisfies the whole bandwidth constraint, there is no need to

balance the traffic load. However, if there are multiple paths the traffic load should be

partitioned optimally for mapping to multiple paths.

[101] Proposed a new constraint based routing algorithm for MPLS networks. It has used

bandwidth & delay constraint. In the algorithm the best path is selected based on best –fit

strategy. In step1, the algorithm eliminates all the links with bandwidth value less than the

bandwidth constraint. In step2, it finds the path with delay value less than (or equal to) delay

constraint value. In step3, if there are more than one feasible path, it selects the path with

minimum hop count. In step 4, the load is distributed & balanced. The best-fit load balancing

strategy selects path from all feasible paths, which has the nearest bandwidth value to the

bandwidth constraint value.

[71] proposed a localized proportional routing approach where each source node collects

information about the traffic originating from itself and computes proportions based solely on

this local information. Global schemes have to gather system wide traffic metrics and thus

slower to react to changes. Localized schemes on the other hand use only local information

Page 7: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

and thus can adapt to change faster. The algorithm has used the strategy, equalization of

blocking probabilities (ebp) of candidate’s paths. This ebp strategy requires only path level

information - the amount of offered load and the corresponding blocking probability. The

objective of ebp strategy is to find a set of proportions such that flow-blocking probabilities

on all the candidate paths are equalized. The strategy can be implemented using the following

procedure to compute new proportions in each iteration. First, the current average blocking

probability is computed. Then the proportion of load onto a path is decreased if it’s current

blocking probabilities are higher than the average and increased if they are lower than the

average. When we talk about path selection, the algorithm selects the candidate paths of a

pair that do not share bottleneck links. In this way of path selection, the blocking probability

can be reduced. To judge the goodness of paths, it has introduced the notion of width for a set

of paths, which is defined as the maximum flow carriable by paths in the set. The amount of

flow carriable by a link is given by its average available bandwidth. So the width of a set of

paths can be computed, given the average available bandwidth information about each link in

the network. Based on the notion of width of a path set, the algorithm proposed a path

selection procedure that adds a new candidate path only if its inclusion increases the width. It

deletes an existing candidate path if its exclusion does not decrease the total width. Therefore,

this proportional routing scheme yields higher through put with lower overhead.

5.3 IMPLEMENTATION OF MULTIPATH APPROACH TO ATTAIN QOS

In this section, a Multi-Path QoS approach for source routing has been proposed.

The proposed algorithm is based on [93]. The two metrics considered in [93] are residual

bandwidth and propagation delay. In [93], the algorithm finds bandwidth-delay constrained

path. As bandwidth is a concave metric and delay is an additive metric, so it finds the path

whose each edge bandwidth satisfies the bandwidth requirement and whose total delay

(addition of delay of all edges) should be less than the maximum delay allowed. The

algorithm first eliminates all the links that do not meet the bandwidth requirement. Then, it

finds the minimum delay path from source to destination using Dijkstra algorithm. The

algorithm is terminated when either destination is reached or delay exceeds the threshold.

The algorithm presented in [93] finds the single best path. We have extended this algorithm

to find multiple paths instead of single path. Our objective is to find link disjoint paths which

satisfy the QoS constraints. To find disjoint paths we have used the ‘Find and Removal’

Page 8: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

method. First, we find the best path between s(source) and t(destination) based on the

metrics and then remove all the links from the network that belong to the path computed, and

then find the second path by running the same algorithm. The algorithm is terminated when

the number of paths reached to m or delay exceeds the threshold or the network is

disconnected from source to destination. Here the value of m is number of adjacent nodes of

source. Thus, the algorithm finds multiple disjoint loopless paths from s to t satisfying QoS

criteria- residual bandwidth, delay in the increasing order of delay.

The basic theme of the algorithm is that, firstly, it eliminates all the links that do not meet the

bandwidth requirement. Then, it finds a (the best one) path satisfying delay. Once a path is

found, all the links in that path are removed from the network graph and the algorithm is run

on the pruned graph to find the next minimum-delay path. The process is continued until the

algorithm comes up with the best m number of shortest paths from s to t or delay exceeds the

delay threshold.

For the construction of paths, we use certain parameters. The following are the list of

parameters used in developing the path construction algorithm.

i. Node structure holds its status (tentative, permanent), its total delay from

source , its predecessor, its bandwidth and number of hops from source

i.e.

1. node. status = Shows the status of node= "permanent" or

"tentative."

2. node. delay= Total delay of path from the source to that node.

3. node. pred = Its previous node according to path found.

4. node. bandwidth= Value of minimum bandwidth edge from

source to that node

5. node. hops=Number of hops from source to that node.

ii. b [i,j] =Residual bandwidth for each edge.

iii. d[i,j] = Delay value for each edge.

iv. band_th =Bandwidth threshold/requirement (Path should have bandwidth

equal or greater than band_th)

v. delay_th = Delay threshold/requirement (Path should have delay equal or

less than delay_th)

vi. count = Number of paths found so far

Page 9: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

vii. m= Number of paths to find= Number of adjacent nodes of source

viii. s= Source node

ix. t= Destination node

Page 10: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

5.3.1 M-Bandwidth-Delay constrained algorithm

The steps of M-Bandwidth-Delay constrained algorithm are-

Step 1: Set dij =0 & bij =0, if bij < Band_th. (Eliminates all links that do not meet the

bandwidth requirement by setting their metric matrix to 0.), count =0.

Step2: Initialize node[s].pred=0, node[s].delay=0, node[s].status=permanent,

node[s].bandwidth=∞.

node[i].pred=s,node[i].delay=∞,node[i].bandwidth=∞,node[i].status=tentative ,for all i ≠s.

k=s

Step 3: for all i whose node[i].flag=tentative && (dki >0)

If (node[i].delay> dki + node[k].delay)

begin

node[i].delay= dki + node[k].delay

node[i].bandwidth=min(node(k).bandwidth, bki)

node[i].pred=k

node[i].hops=node[k].hops+1

end

End if

Step 4 Check for connectivity of source & destination if false then exit.

Step 5: Change k so that node[k].delay = Min (node[i].delay) (Here i is tentative node)

node[k].flag=permanent

Step 6 :Check

If k= t then

begin

If (node[k].delay<=delay_th) then

begin

count=count+1

record the path by following predecessor link.

If ( count <=m) then

Set bij, dij=0 for all edges available in path found & Goto step 2

Else

Page 11: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

Stop

End if

End

Else

Stop

End if

End

Else

Go to Step 3

End if

Page 12: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

YesYesYesNoAmong the tentative nodes, find the node having least delay value & make it as active node. Mark active as permanent StartEliminate all links that do not meet the bandwidth requirement by making its entry in b[ij] and d[ij] as zero, count=0

Identify source and destination node as s and t, Initialize the parameters of the node structure record (for all nodes) and set their status to "tentative" and their delay, bandwidth to "infinity"

Make source as active node and set its bandwidth to ∞ and delay to 0 & make source as permanent . active=tcount=count+1,Record the path by tracing the predecessor node . Subtract every entry in b[i,j] and d[ij] that appears in the path

Update all tentative nodes and record its structure i.e. its delay , its bandwidth ,hops ,its predecessor according to active node.

delay <= delay_thStopNoYe sStop count= m

Check Connectivity

NoStopNo

Page 13: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

5.3.2

Illustration of Algorithm

We have illustrated the algorithm through the following example-

Here is a network of 6 nodes. Each edge is represented by its residual bandwidth & delay

metric

3/34/55/55/35/25/34/25/34/35/4124536

Figure 5.1- Flowchart of M-Bandwidth-Delay constrained Algorithm

Page 14: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

Figure 5.2 Example Network with edges depicting bandwidth/delay

Thus bandwidth and delay matrix of graph are-

Table 5.1- Bandwidth Matrix

1 2 3 4 5 6

1 0 5 5 4 0 02 5 0 0 4 4 0

3 5 0 0 5 0 5

4 4 4 5 0 3 55 0 4 0 3 0 5

6 0 0 5 5 5 0

Table 5.2 –Delay Matrix

1 2 3 4 5 6

1 0 3 2 5 0 02 3 0 0 2 3 0

3 2 0 0 4 0 5

4 5 2 4 0 3 35 0 3 0 3 0 3

6 0 0 5 3 3 0

The following parameters have been assumed-

Bandwidth threshold - 4

Delay threshold - 10

Page 15: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

Source - 1

Destination – 6

Total number of paths found-Adjacent edges of Source - 3

The problem can be defined as - find multiple disjoint paths between 1 and 6 in the given

network. The paths should follow the following conditions-

The bandwidth of the paths should be minimum 4.

The delay of the paths should not exceed 10.

As bandwidth is concave metric, so each of the edges in the path found should have value

greater than or equal to 4 and delay is an additive metric, so the sum of the delay of edges

constituting the path should be less than or equal to 10.

According to the algorithm, the edges are removed from the graph that does not satisfy the

bandwidth requirement by setting their delay value and bandwidth value to 0.

As the bandwidth requirement is 4 so the edges having less than bandwidth 4 should be

removed from the graph. Among all the edges of graph, the edge 4-5 is having bandwidth 3

i.e. less than 4. It will not be considered further for path finding.

Now the resultant network will be-

4/55/55/35/25/34/25/34/35/4124536

Page 16: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

Figure 5.3 Filtered Network

In this network, Elimination of edges not satisfying bandwidth requirement has been done.

Now the bandwidth of all the edges is equal or greater than 4.

First Iteration

In the network depicted in fig. 5.3, the shortest delay path from 1 to 6 as computed by the

proposed algorithm is

Table 5.3 –Path Table

Path Bandwidth of Path Delay of Path Hop-count1-3-6 5 7 2

Remove the edges found in this path from the network.

The resultant network will be

4/55/35/34/25/34/35/4124536

Page 17: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

Figure 5.4 – Network after removal of edges of first path

Second Iteration

Now in the network of figure 5.4, the least delay path from 1 to 6 is

Table 5.4 –Path Table

Path Bandwidth of Path Delay of Path Hop-count1-4-6 4 8 2

Remove the edges found in this path from the network.

The resultant network will be

5/45/34/25/34/3124536

.

Figure 5.5-Network after removal of edges of second path

Third iteration

Now in network of figure 5.5, the least delay path from 1 to 6 is

Table 5.5 –Path Table

Page 18: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

Path Bandwidth of Path Delay of Path Hop-count1-2-5-6 4 9 3

Here the value of total number of paths found has reached to 3.So the algorithm will

terminate here.

Thus, the multiple disjoint QoS paths found in the given network are as follows-

Table 5.6 –Path Table

Path Bandwidth of Path Delay of Path Hop-Count1-3-6 5 7 21-4-6 4 8 21-2-5-6 4 9 3

Thus the algorithm finds multiple disjoint paths satisfying bandwidth and delay requirements

in the order of increasing delay. This algorithm has been implemented in MATLAB and

presented in Appendix A. The output obtained in MATLAB has been presented in next

section.

Page 19: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

5.3.3 MATLAB Output

Input number of nodes:-6

Enter bandwidth and delay matrix:-

Enter bandwidth between [1] [1] 0

Enter delay between [1 ][1 ]0

Enter bandwidth between [1] [2] 5

Enter delay between [1 ][2 ]3

Enter bandwidth between [1] [3] 5

Enter delay between [1 ][3 ]2

Enter bandwidth between [1] [4] 4

Enter delay between [1 ][4 ]5

Enter bandwidth between [1] [5] 0

Enter delay between [1 ][5 ]0

Enter bandwidth between [1] [6] 0

Enter delay between [1 ][6 ]0

Enter bandwidth between [2] [1] 5

Enter delay between [2 ][1 ]3

Enter bandwidth between [2] [2] 0

Enter delay between [2 ][2 ]0

Enter bandwidth between [2] [3] 0

Enter delay between [2 ][3 ]0

Enter bandwidth between [2] [4] 4

Enter delay between [2 ][4 ]2

Enter bandwidth between [2] [5] 4

Enter delay between [2 ][5 ]3

Enter bandwidth between [2] [6] 0

Enter delay between [2 ][6 ]0

Enter bandwidth between [3] [1] 5

Enter delay between [3 ][1 ]2

Enter bandwidth between [3] [2] 0

Enter delay between [3 ][2 ]0

Enter bandwidth between [3] [3] 0

Enter delay between [3 ][3 ]0

Enter bandwidth between [3] [4] 5

Enter delay between [3 ][4 ]4

Enter bandwidth between [3] [5] 0

Page 20: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

Enter delay between [3 ][5 ]0

Enter bandwidth between [3] [6] 5

Enter delay between [3 ][6 ]5

Enter bandwidth between [4] [1] 4

Enter delay between [4 ][1 ]5

Enter bandwidth between [4] [2] 4

Enter delay between [4 ][2 ]2

Enter bandwidth between [4] [3] 5

Enter delay between [4 ][3 ]4

Enter bandwidth between [4] [4] 0

Enter delay between [4 ][4 ]0

Enter bandwidth between [4] [5] 3

Enter delay between [4 ][5 ]3

Enter bandwidth between [4] [6] 5

Enter delay between [4 ][6 ]3

Enter bandwidth between [5] [1] 0

Enter delay between [5 ][1 ]0

Enter bandwidth between [5] [2] 4

Enter delay between [5 ][2 ]3

Enter bandwidth between [5] [3] 0

Enter delay between [5 ][3 ]0

Enter bandwidth between [5] [4] 3

Enter delay between [5 ][4 ]3

Enter bandwidth between [5] [5] 0

Enter delay between [5 ][5 ]0

Enter bandwidth between [5] [6] 5

Enter delay between [5 ][6 ]3

Enter bandwidth between [6] [1] 0

Enter delay between [6 ][1 ]0

Enter bandwidth between [6] [2] 0

Enter delay between [6 ][2 ]0

Enter bandwidth between [6] [3] 5

Enter delay between [6 ][3 ]5

Enter bandwidth between [6] [4] 5

Enter delay between [6 ][4 ]3

Enter bandwidth between [6] [5] 5

Enter delay between [6 ][5 ]3

Enter bandwidth between [6] [6] 0

Page 21: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

Enter delay between [6 ][6 ]0

The bandwidth matrix is :

0 5 5 4 0 05 0 0 4 4 05 0 0 5 0 54 4 5 0 3 50 4 0 3 0 50 0 5 5 5 0

The delay matrix is:

0 3 2 5 0 03 0 0 2 3 02 0 0 4 0 55 2 4 0 3 30 3 0 3 0 30 0 5 3 3 0

Enter the source:-1

Enter the destination:-6

Enter delay threshold:-10

Enter bandwidth threshold:-4

The 1 path between 1 to 6 is 1 - 3 - 6

The total delay of path is- 7

The number of hops - 2

The bandwidth of path - 5

The 2 path between 1 to 6 is 1 - 4 - 6

The total delay of path is- 8

The number of hops - 2

The bandwidth of path - 4

The 3 path between 1 to 6 is 1 - 2 - 5 - 6

The total delay of path is- 9

The number of hops - 3

The bandwidth of path - 4

Page 22: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

Figure

5.6 Path Diagram

In this MATLAB output network-

The edge values are representing the delay value.

Node 1, Node 6 are - source, destination.

Red Edges are representing- 1st path

Blue Edges are representing- 2nd path

Green edges are representing- 3rd path

Page 23: CHAPTER 5 MINGLING MULTIPATH WITH QUALITY OF SERVICEshodhganga.inflibnet.ac.in/bitstream/10603/39874/12/12_chapter 5.p… · Ford algorithm is single QoS path computation algorithm

5.4 SUMMARY

The areas that have been studied to merge QoS and multipath are- Dijkstra algorithm,

Bellman-Ford algorithm, Resource reservation & MPLS.

[93] has described an algorithm based on Dijkstra for single path. A multi-path extension to

this algorithm has been proposed in this chapter with an illustration through an example

network. The QoS parameters that have been considered are bandwidth and delay. The worst

case complexity of the algorithm in [93] is O(n2). Thus the complexity of the proposed

algorithm is O (m.n2). Here n represents the number of nodes and m represents the number of

paths computed. This is a source routing algorithm. As the main objective of the present

research work is to define a fully distributed approach being aware of its advantages and

importance, distributed approach has been explored in next chapter.