,(((,QWHUQDWLRQDO&RQIHUHQFHRQ,QIRUPDWLRQ&RPPXQLFDWLRQ,QVWUXPHQWDWLRQDQG&RQWURO,&,&,&ದ

3DSHU,G

Ordering Policy based Routing and Bandwidth

Assignment Algorithms in Optical Networks
Deepak Batham1, Anmol Jain2, Pritesh Gethewale2, Yogesh Kherajani2, Umang Gupta2
1
Department of Electronics & Instrumentation Engineering
2
Department of Electronics & Telecommunication Engineering
1,2
Institute of Engineering & Technology, Devi Ahilya University, Indore, 452001, India
Email: {deepakbatham786, anmoljainaj10, priteshg228, kherajaniyogesh, umanggupta.iet}@gmail.com

Abstract— Today, the Internet traffic is crowded by emerging high and low. The services which require high bandwidth are
bandwidth hungry multimedia services. These services video-on-demand, teleconferencing, etc. On the other hand,
required dynamic bandwidth which may be high, low or the services which require low bandwidth are E-mail, text-
moderate. For establishing such a connection demand, we chat, etc. Thus, the optical networks play an important role in
propose a novel routing and bandwidth assignment (RBA) wide variety of fields like digitization, encryption, decryption,
algorithm. In order to efficiently utilize the network resources and cryptography, where we need varying bandwidth.
and to reduce the connection blocking probability, we propose The Indian government has also emphasized on
various ordering policies based RBA algorithms. For RBA, digitization. The vision behind the Digital India project is to
two constraints, the wavelength continuity and wavelength empower every citizen with access to digital services,
contiguity constraint must be satisfied, which is different and knowledge and information. Thus, the Internet connection has
complex with the traditional routing and wavelength become quite important in government/private offices,
assignment (RWA) technique. In ordering policies based RBA hospitals, schools, universities, industries, gram panchayats,
algorithms, the entire set of traffic demands are arranged etc. To achieve this target, National optical Fiber Network
either by their route length, or bandwidth requirement, or (NOFN) program begins which is one of the most important
considering both. pillars of Digital India project, has aimed to connect 2.5 lakh
We performed the extensive simulation experiments in gram panchayats.
MATLAB environment and compared the proposed policies To provision such a connection request in optical
with the simple RBA algorithm. It is observed from the result networks, the problem of route searching and allocating
that some of the proposed ordering policies outperform the wavelength to each link of the selected route is referred as
RBA, on the metric of blocking probability, and resource routing and wavelength assignment (RWA) problem [3-8].
utilization ratio. The connection request is represented by a tuple of three
elements (s, d, n) where,‘s’ is the source node, ‘d’ is the
Keywords—routing and bandwidth assignment, ordering policy, destination nodes, and ‘n’ is the number of requested
optical networks, blocking probability, resource utilization ratio wavelength. In conventional RWA, requested wavelength is
one (i.e., n = 1) and is same for all connection requests. But,
for bandwidth intensive services, requested wavelength is
I. INTRODUCTION
varying or dynamic. In this paper, we consider four types of
In last few years, the spread of Internet has dramatically bandwidth services where, n=1, 2, 3, and 4. The requested
increased the number of end users who, at the same time, are wavelength is also called requested bandwidth.
more demanding, in terms of bandwidth capacity The provisioning of different bandwidth intensive services
consumption. To address the bandwidth limitation of current with efficient resource utilization seems impossible with the
electronic network infrastructures and to cope with the rapid currently deployed RWA algorithm. Thus, a novel routing and
growth of Internet traffic and bandwidth-on-demand services, bandwidth assignment (RBA) algorithm is introduced. The
optical wavelength division multiplexing (WDM) problem of route searching and allocation of desired number
communication system have been deployed in many of wavelengths (or requested wavelength) to the selected route
telecommunication backbone networks [1]. This has is known as RBA. Similar to RWA, the RBA problem is NP-
significantly increased the bandwidth capacity, data Hard. So, it can be solved separately as the routing
transmission rate, range of network, and has reduced the cost. subproblem and the bandwidth allocation sub problem. The
The Internet traffic consist of bandwidth intensive services ASON and GMPLS switching and control plane supports
such as multimedia, cloud computing, grid computing, etc [2]. bandwidth varying demand [9-10].
These services have different bandwidth requirements i.e.,
,(((,QWHUQDWLRQDO&RQIHUHQFHRQ,QIRUPDWLRQ&RPPXQLFDWLRQ,QVWUXPHQWDWLRQDQG&RQWURO,&,&,&ದ
3DSHU,G

In RWA, same wavelength must be used in all links along end free wavelength in optical networks for future connection
the candidate route, known as wavelength continuity requests. In [6], authors proposed greedy RWA algorithm for
constraint (WCC). The RBA is different from and more multi-domain optical networks under scheduled traffic.
challenging than the traditional RWA problem. In addition to Authors in [8] proposed the least weight wavelength
WCC, a wavelength contiguity constraint is also considered. assignment algorithms for static and dynamic RWA problem,
We are assuming here, non-wavelength convertible network in order to reduce the BP. A priority based RWA scheme has
means no wavelength conversion take place at any node in the
been proposed with the incorporation of traffic grooming
network. The wavelength contiguity constraint implies that the
mechanism in [9], to reduce connection blocking. In [10],
requested wavelength must be consecutive.
Thus, RBA for varying bandwidth demands is an authors proposed an improved wavelength assignment
important issue. For any network service provider, the aim is algorithm named as most-used wavelength-conversion
to efficiently utilize the network resources with minimum (MUWC). In this algorithm, MU wavelength assignment
connection blocking. Lower the connection blocking means algorithm is executed until connection blocking. When the
higher is the number of connection establishment. Therefore, connection is blocked, wavelength conversion is introduced
higher revenue is generated for the network service provider. and hence BP gets reduced.
We proposed a novel simple RBA algorithm (without ordering To the best of our knowledge, no such footprints are
policy) and various ordering policies based RBA algorithms. available for RBA problem. To provision the bandwidth
The simulation results of some proposed policy based RBA intensive traffic in optical networks we proposed the ordering
strategies have significantly improved the performance in term policies based RBA algorithms with the aim is to reduce the
of the blocking probability (BP) and resource utilization ratio BP and RUR.
(RUR).
The rest of the paper is organized as follows: Section-II
deals with the related work. Section-III represents the detailed III. EXISTING AND PROPOSED STRATEGIES:
description of simple RBA, and various ordering policy based
RBA strategies. Section-IV illustrates the simulation results 1. Routing and Bandwidth Assignment (RBA):
and analysis. Conclusion of the research and the direction for In RBA, demands are arranged according to first come
the future work are reported in Section V. first serve (FCFS) policy. The demand arrived earlier will be
served first. After searching a candidate route from K-alternate
route, check the two RBA constrains i.e., wavelength
II. RELATED WORK contiguity and continuity constraint on the selected route, if
The present section deals the related work based on RWA. free resources (wavelengths) are available as per the demand
In optical networks, the traffic is classified as: (i) static [3-4], requirement, it will be assigned to that route and accept the
(ii) dynamic [5], and (iii) scheduled traffic [6-7]. If the entire connection request.
set of connection request is known in advance, such traffic is
2. Lowest Bandwidth Demand First (LBDF)
known as static traffic. In dynamic traffic, the connection
requests are arrived randomly. In scheduled traffic, the setup In LBDF, the demands are arranged in ascending order of
and tear-down time of the connection requests/demands are their requested bandwidth. That is, the demands which require
known in advance. In [3], authors proposed path length based lowest bandwidth will be served first.
RWA strategy for static traffic. The simulation results showed 3. Highest Bandwidth Demand First (HBDF)
reduction in blocking probability with the use of lesser number
In HBDF, the demands are arranged in descending order of
of wavelength links in the network.
their requested bandwidth. That is, the demands which require
In [4], authors review various routing strategies and
highest bandwidth will be served first.
evaluate their performance on the metric of blocking
probability under several wavelength allocation strategies such 4. Shortest Path Demand First (SPDF)
as Random-Fit (RF), First-Fit (FF), Most Used (MU), and In SPDF, the demands are arranged in increasing order of
Least Used (LU). RF, checks wavelengths randomly for the their route length. The demands which require minimum
selected route and assigns a free wavelength in all links of the number of hops on the route will be served first.
route. FF strategy checks, the wavelengths in prearranged 5. Longest Path Demand First (LPDF)
order to do the same. In MU wavelength assignment strategy,
In LPDF, the demands are arranged in decreasing order of
first updates the wavelengths in descending order of the usage
their route length. Larger the route length means larger is the
and then checks the sorted wavelengths to do the same. In LU number of hops. The demands which require largest hops
assignment strategy, first updates the wavelengths according route will be served first.
to the ascending order of the usage and then checks the sorted
wavelengths to do the same. It has been demonstrated that the 6. Lowest Bandwidth Shortest Path Demand First (LBSPDF)
FF outperforms and is extensively utilized because of its In LBSPDF, the demands are arranged in ascending order
inherent simplicity and favorable capacity of leaving end-to- of their bandwidth as well as number of hops of the route.
,(((,QWHUQDWLRQDO&RQIHUHQFHRQ,QIRUPDWLRQ&RPPXQLFDWLRQ,QVWUXPHQWDWLRQDQG&RQWURO,&,&,&ದ
3DSHU,G

That is the demands have lowest bandwidth and shortest route

will be served first.
7. Lowest Bandwidth Longest Path Demand First (LBLPDF)
In LBLPDF, the demands are arranged in ascending order
of their bandwidth and descending order of their hop count.
That is, the demands have lowest bandwidth and longest route
will be served first.
8. Highest Bandwidth Shortest Path Demand First
(HBSPDF)
In HBSPDF, the demands are arranged in decreasing order
of their bandwidth and increasing order of their hop count. Fig. 2. ARPANET Topology (20 Nodes, 32 Links)
The demands have highest bandwidth and shortest route will
be served first. The blocking probability (BP) and resource utilization
ratio (RUR) are considered as output parameters. BP refers to
9. Highest Bandwidth Longest Path Demand First the ratio of number of blocked connection demand to the
(HBLPDF) number of arrived demands in the network. Lower the BP,
In HBLPDF, the demands are arranged in decreasing order higher is the connections established. Thus, for a network
of their bandwidth as well as hop count. The demands have service provider higher the number of connections established,
highest bandwidth and longest route will be served first. more is the revenue generated. Another parameter of interest
to describe how efficiently, the network resources are used
i.e., RUR. RUR is defined as the ratio of network capacity
IV. SIMULATION RESULTS: (wavelength links) used to the number of established demand
This section presents the simulation results and discussion. in the network.
In order to implement the proposed ordering policies based Fig. 3 and 4 shows, the plot of BP versus the number of
RBA and to compare their performances with simple RBA demands arrived for US network and ARPANET topologies,
algorithm, a simulation model has been developed in respectively. The BP of the simple and ordering policies based
MATLAB environment. The simulation results are tested on RBA strategies, increase with the increase in the number of
the standard network topologies, US network and ARPANET arrived demands. It is evident from the figures that the BP of
as shown in Fig. 1 and 2, respectively. The US network HBLPDF, HBDF, LPDF, and LBLPDF ordering policies are
topology comprises of 15 nodes and 25 links and the higher than the simple RBA strategy because it seems to be
ARPANET topology consists of 20 nodes and 32 links. The arrange the whole demand set in such a manner that the
information regarding the network topologies are separately longest route length and highest bandwidth demand first.
feed as input parameters in the simulation model. We assume Thus, it consumes most of the resources for lower load. The
each link is bidirectional and have 20 wavelengths per link in HBSPDF, SPDF, LBDF, and LBSPDF show lower BP as
the network. A total of 300 connection requests/demands have compare to RBA strategy because of the shortest route and
been considered for both network topologies to stimulate the lowest bandwidth demands are considered as higher priority.
results. Thus, for lower load network resources are underutilized.
Hence, are available for future demands or higher load.
Among all the ordering policies, LBSPDF outperforms for
both network topologies. LBSPDF shows 44.04% and 42.47%
reduction in BP for US network and ARPANET as compare to
RBA.
The next parameter of interest is the RUR. Fig. 5 and 6,
show the plot of RUR versus the number of demands arrived
for US network and ARPANET topologies, respectively.
Lower the value of RUR means resources are utilized
efficiently. This is shown in figure that RUR of HBSPDF,
SPDF, LBDF, and LBSPDF show momentous improvement
as compared to the other ordering policies and RBA. Among
all the strategies, LBSPDF shows best result. LBSPDF shows
46.30 % and 50.65% improvement in the value of RUR as
Fig. 1. US Network Topology (15 Nodes, 25 Links) compare to the simple RBA strategy on US network and
ARPANET, respectively.
,(((,QWHUQDWLRQDO&RQIHUHQFHRQ,QIRUPDWLRQ&RPPXQLFDWLRQ,QVWUXPHQWDWLRQDQG&RQWURO,&,&,&ದ
3DSHU,G

Fig. 3. BP Vs Number of Demands Arrived for US Network Topology

Fig. 6. RUR Vs Number of Demands Arrived for ARPANET Topology

V. CONCLUSION
The paper evaluates the impact of ordering policies based
RBA algorithms in optical networks. The simulation results of
the proposed SPDF, LBDF, and LBSPDF are superior as
compare to the simple RBA strategy in terms of BP and RUR.
Among all, LBSPDF outperforms. LBSPDF also generate
highest revenue for the network service provider with the use
of minimum network resources. LBSPDF shows 44.04% and
42.47% reduction in BP for both US network and ARPANET,
respectively as compare to simple RBA algorithm. Similarly
46.30% and 50.65% improvement in the value of RUR is
recorded as compare to the simple RBA on US network and
ARPANET topology, respectively.

References
[1] Y. Coulibaly, G. Rouskas, M.S.A. Latiff, M.A. Razzaque, and S.
Fig. 4. BP Vs Number of Demands Arrived for ARPANET Topology Mandala, “QoS-aware ant-based route, wavelength and timeslot
assignment alorithm for optical burst switched networks,” Trans.
Emerging Tel. Tech., 2015. DOI: 10.1002/ett.2919.
[2] D. Batham, D.S. Yadav, and S. Prakash, “ Efficient resource
provisioning using traffic balancing in multidomain optical networks,”
Int. J. Commun. Syst., 2016. DOI: 10.1002/dac.3203.
[3] D. Batham, S. Nikara, S. Gupta, and V. Bagdi, “Static path-length based
traffic routing and wavelength assignment strategy in WDM optical
transport networks,” National Conference on RAMCTM. Indore India,
2013.
[4] H. Zang, J.P. Jue, and B. Mukherjee, “Review of routing and
wavelength assignment approaches for wavelength routed optical WDM
networks,” Optical Networks Magazine, vol. 1, pp. 47-60, 2000.
[5] X. Chu, and B. Li, “Dynamic routing and wavelength assignment in the
presence of wavelength conversion for all-optical networks,”
IEEE/ACM Transactions on Networking, vol. 13, pp. 704-715, JUNE
2005.
[6] D. Batham, D.S. Yadav, and S.Prakash, “Greedy RWA algorithm for
scheduled traffic in multi-domain optical networks,” in Proc. of 12th
International Conference on Fiber Optics and Photonics (PHOTONICS-
2014), IIT Kharagpur, Dec. 2014, pp. 1-3. doi:
10.1364/PHOTONICS.2014.T3A.53
[7] D. Batham, M. Rajvedya, D.S. Yadav, S. Prakash, “Wavelength
Fig. 5. RUR Vs Number of Demands Arrived for US Network Topology
Indexing based End-to-end Routing in Multi-Domain Optical Networks
under Scheduled Lightpath Demand,” Proc. IET International
,(((,QWHUQDWLRQDO&RQIHUHQFHRQ,QIRUPDWLRQ&RPPXQLFDWLRQ,QVWUXPHQWDWLRQDQG&RQWURO,&,&,&ದ
3DSHU,G

Conference on Signal Processing (ICSP-2016), Vidisha, India, Nov. [9] B.C. Chatterjee, N. Sarma, and P.P. Sahu, “Priority based routing and
2016. wavelength assignment with traffic grooming for optical networks,” J.
[8] R. Randhawa, and J.S. Sohal, “Static and dynamic routing and Opt. Commun. Netw., vol. 4, pp. 480-489, 2012
wavelength assignment algorithms for future transport networks,” Optik, [10] A. Wason, and R.S. Kaler, “Wavelength assignment algorithms for
vol. 121, pp. 702–710, 2010. WDM optical networks,” Optik, vol. 122, pp. 877–880, 2011.