A Novel Hop-distance Sensitive Approach to Elastic Optical Networks

RSA Algorithms

Stathis B. Mavridopoulos

1 a

, Georgia A. Beletsioti

1 b

, Georgios A. Tziroglou

1 c

Constantine A. Kyriakopoulos

1 d

, Petros Nicopolitidis

1 e

, Georgios I. Papadimitriou

1 f

and Emmanouel Varvarigos

2 g

Department of Informatics, Aristotle University of Thessaloniki, Thessaloniki, GR-54124, Greece

School of Electrical and Computer Engineering, National Technical University of Athens, Athens, GR-15780, Greece

Keywords: Elastic Optical Networks, Hop Distance, Fairness, Normalized Blocking Probability, Backbone, Metropolitan.

Abstract:

Elastic optical networks (EON) allow great ﬂexibility through ﬁner spectrum allocation granularity when

compared to traditional WDM solutions. Their improved spectrum efﬁciency makes them a promising solution

for next-generation backbone and metropolitan networks. Distant connections in elastic optical networks that

are routed through multiple hops suffer from increased bandwidth blocking probability (BBP), in contrast to

easier formulation of more direct connections. Traditional BBP as a metric fails to capture this phenomenon.

In this work, a normalization of BP to the connection’s hop distance is proposed and a novel low complexity

algorithm is presented that takes this new metric into consideration. Simulation results show that the proposed

scheme improves network performance and fairness with no deterioration of BBP, when compared to the

FirstFit RSA algorithm.

1 INTRODUCTION

Globally, the IP trafﬁc in telecommunication net-

works has risen at unprecedented levels, expanding

annually at an average rate of 24%, according to Cisco

Global - 2021 Forecast Highlights (Cisco, 2016). Not

only do the number of users increase, but also new,

bandwidth starving applications such as voice over IP,

video on demand, high- deﬁnition video and gaming

(Beletsioti et al., 2016).

Traditionally, optical technologies based on

Wavelength Division Multiplexing (WDM) (e.g (Kyr-

iakopoulos et al., 2018b), (Beletsioti et al., 2018)),

which use the conventional ﬁxed grid of 50 GHz and

100 GHz for backbone and metro networks (Sim-

mons, 2014) respectively, have been used to accom-

modate the requested trafﬁc demands. However, to-

day’s extremely high trafﬁc needs threaten to make

https://orcid.org/0000-0002-7058-3147

https://orcid.org/0000-0002-1895-094X

https://orcid.org/0000-0002-5771-1511

https://orcid.org/0000-0001-7874-2205

https://orcid.org/0000-0002-5059-3145

https://orcid.org/0000-0001-9529-9380

https://orcid.org/0000-0002-4942-1362

the traditional WDM technologies inadequate. Never-

theless, Elastic Optical Networks (EON) in conjunc-

tion with OFDM scheme is assumed to be a promis-

ing technology coping with the increasing and more

diverse trafﬁc demands (Gerstel et al., 2012), (Kyri-

akopoulos et al., 2018a).

EONs offer ﬂexibility in how to allocate just the

appropriate bandwidth capacity to the connections

and are considered the network solution for back-

bone and next generation metropolitan networks. In

ﬂexible grid technologies the spectrum is split into

slots of 6.75 GHz, 12.5 GHz or 25 GHz. These fre-

quency slots (FS) are then combined to create chan-

nels, which are not overlapping due to OFDM’s or-

thogonality condition (Vizca

ıno et al., 2012), in order

to serve custom sized bandwidth requests and adapt

to the dynamic and heterogeneous nature of their ar-

rivals during network’s operation.

The main focus of routing and spectrum alloca-

tion assignment EON algorithms (RSA) is that con-

nections are required to satisfy both the continuity and

contiguity frequency constraints (Chatterjee et al.,

2018). Under high trafﬁc loads, requests between dis-

tant nodes may suffer increased blocking rates due

to the fact that these constraints need to be satisﬁed

Mavridopoulos, S., Beletsioti, G., Tziroglou, G., Kyriakopoulos, C., Nicopolitidis, P., Papadimitriou, G. and Varvarigos, E.

A Novel Hop-distance Sensitive Approach to Elastic Optical Networks RSA Algorithms.

DOI: 10.5220/0007926900710077

In Proceedings of the 16th International Joint Conference on e-Business and Telecommunications (ICETE 2019), pages 71-77

ISBN: 978-989-758-378-0

across multiple links.

Our work demonstrates the fairness problem that

multihop connections face. Satisfying the continu-

ity and contiguity constraint over longer distances be-

comes increasingly harder, when compared to estab-

lishing connections comprising of only a few hops.

To this end, we propose a new metric in this paper, the

normalize to hop-distance Bandwidth Blocking Prob-

ablity (normalized BBP), that takes connection hop-

distance into consideration and its effect on network

performance.

A similar fairness issue is highlighted in (Rosa

et al., 2015), where it is shown that bandwidth

consuming connections are characterized by higher

blocking probability. Finding larger continuous

groups of unoccupied FSs is hard; contiguity exac-

erbates the situation. Light bandwidth requests are

more probable to ﬁt in the spectrum and experience

lower probability of blockage.

Another work that takes interest into connection

distance is (Chatterjee and Oki, 2016). Low index FSs

are affected by dispersion less than high index ones,

meaning that long distance lightpaths suffer more

quality-of-transmission degradation when placed in a

high index FS. By taking this phenomenon into con-

sideration, the number of FSs per connection can be

minimized, leading to improvements for the network

performance. Our work takes a different approach

since it utilizes connection hop-distance instead of the

actual lightpath length.

A low complexity, effective RSA solution that is

used in many works is the FirstFit (FF) algorithm

(Zheng et al., 2010), (Ning and Shen, 2012). Our pro-

posed method, called HopWindows (HW), improves

upon FF by taking hop-distance into consideration.

In Section 2 the motivation for this work is further

explained. Section 3 provides a detailed description

on the operations of the HW algorithm. Finally, in

Section 4 the performance evaluation of our method

is presented. It is shown, that the proposed method

can achieve improvements from 5% up to 36% when

compared to FF against the normalized BBP metric,

for a variety of underlying topologies (metro-mesh,

backbone and metro-ring).

2 MOTIVATION

One of the most commonly used metrics employed

when investigating the performance of elastic opti-

cal networks is bandwidth blocking probability (BBP)

and is calculated as:

Figure 1: Optical backbone network NSFnet, with 14 nodes

and 21 links (Shen and Tucker, 2009).

bbp =

∑

bandwidth blocked

∑

total bandwidth generated

(1)

This metric is undeniably useful, but doesn’t re-

ﬂect the total reality, especially concerning connec-

tions comprising many hops. Depending on the un-

derlying routing algorithm, one or more paths are

available betweeen a pair of nodes. Node connections

that can only be established over more than one inter-

mediate nodes are called high-hop connections. Node

pairs that are neighbors or require only one other in-

termediate nodes are low-hop connections. Minimum

hop distance (conn min hops) is considered in our

method and refers to the minimum hop length of all

allowed paths between two nodes. Neighbor nodes

have a minimum hop distance of one.

Elastic optical networks allow great ﬂexibility and

increased bandwidth utilization, however connections

are required to satisfy both the continuity and contigu-

ity frequency constraints. Since high-hop connections

require free overlapping frequency slots over multiple

links, they suffer from higher blocking probability.

This phenomenon can be observed when BBP is

analyzed per minimum connection hop distance. In-

dicatively, in our simulation of First Fit performance

for the NSFnet topology in Section 4 (Figures 3

and 5), average BBP of all connections for 150 Er-

langs may be 0.06, however BBP for connection with

conn min hops = 1 and 2 is 0.001 and 0.0487 respec-

tively, while BBP exponentially increases in the cases

of conn min hops = 3 and 4, to 0.10 and 0.215 re-

spectively.

3 PROPOSED SCHEME

3.1 HopWindows Scheme

The proposed HopWindows (HW) algorithm is de-

scribed in Algorithm 1. Similarly to the FirstFit

method (Zheng et al., 2010), when a new connection

request is generated, the k shortest paths between

DCNET 2019 - 10th International Conference on Data Communication Networking

Algorithm 1: HopWindows.

1 New Connection

Input : conn min hops: minimum number

of hops for this connection

Paths[i]: k3 shortest paths

Paths[i][max util]: Percentage of

FSs in path that are occupied

Paths[i][len]: number of hops for

this path

Paths[i][ f ree slots]: available FSs

on path

masks[conn min hops]: limit FS

according to the mask

2 foreach path ∈ Paths do

3  Find a path that ﬁts this connection

4 if path[max util] > 50% then

5  This path is at risk of congestion

6 if path[len] > conn min hops + 1

then

7 continue

8 else

9 apply masks[conn min hops] on

path[ f ree slots]

10 if connection ﬁts in path[ f ree slots] then

11  Apply ﬁrstﬁt for this path

12 return path

Output: Select Path | Reject connection

source and destination are examined for unoccupied

FSs, that are continuous and contiguous. Those free

FSs are stored in the array path[ f ree slots].

The critical step of HW is Line 4, where the pro-

posed method is differentiated from the simple FF

method. In the case of HW, the maximum path uti-

lization is calculated. Each link in that path will have

a different degree of utilization, and maximum path

utilization refers to the utilization of the most “ﬁlled”

link in that path. If maximum path utilization is higher

than a threshold, then that path contains a link that is

either a network bottleneck, or simply, the network

offered load is sufﬁciently high and that path is al-

ready congested. Our investigation suggested that us-

ing 50% for that threshold produces good results.

The next step, when HW mode is enabled for that

connection path, is to examine if the current path is

longer than the shortest in hops from all paths. This

check preemptively disables connections from estab-

lishing “circular” paths and is integral part of our pro-

posed method. A side beneﬁt of this rule, is that con-

nections may select a longer but less occupied path,

Figure 2: Mesh based metropolitan network with 29 nodes

and 41 links (Antoniades et al., 2004).

resulting in a simple load balancing of the network

links.

If the connection is allowed to proceed to Step 9,

then a mask is applied to the path[ f ree slots]. The

form of the mask depends on the minimum hop dis-

tance of the source - destination pair, thus a different

mask is enforced in the case of neighboring nodes,

than in the case of N hop-distant nodes. The en-

forcement of this mask further limits the allowed FSs

for this connection. For example, the meaning of

a mask like mask[1] = [0, 100], is that neighboring

nodes (with hop-distance = 1) are allowed to only use

FSs with index from 0 to 100, disallowing FSs with

index larger than 100. Practically, this limitation may

force some 1-hop connections to be rejected, but re-

serves bandwidth for high hop connects that are un-

derperforming.

After the mask rule is applied to the

path[ f ree slots], the algorithm attempts to place the

connection to the leftmost slot of continuous FSs

that can ﬁt it, similarly to how FF operates. If the

connection fails to ﬁt, the algorithm examines the

next available path or the connection is denied. In

that case the connection is “blocked”.

3.2 Selecting the Hop-distance Masks

A critical component of our method is the form of

the masks utilized in the proposed method. Our

analysis through simulation demonstrates that masks

are unique to each topology employed. There is no

generic rule set that can satisfy multiple and diverse

scenarios, such as those examined in Section 4. Our

suggestion to this problem is to diverge to a good so-

lution by pre-processing.

In order to ﬁnd a good mask for our proposed

method, a new metric is introduced that our diver-

gence technique attempts to reduce, which is called

A Novel Hop-distance Sensitive Approach to Elastic Optical Networks RSA Algorithms

Figure 3: Blocking probability per connection hop distance.

For each load value, the left bar represents FirstFit and right

the HW method. BBP is represented as bar high. Lower

sections of the bars represent the blocking probability of

low hop-distance connections. HW improves upon BBP,

while the largest beneﬁt is on high hop-distance connec-

tions.

Figure 4: Total network trafﬁc load analysis. The dotted bar

represents the FF successful connections, the slashed bar

the HW successful connections, bars with horizontal lines

the FF blocked connections and the crossed bar the HW

blocked connections. Lower sections of the bars represent

the trafﬁc of lower hop-distance connections. HW achieves

more successful 3-hop and 4-hop connections.

normalized bpp and is calculated as:

normalized bbp =

∑

bandwidth blocked · min hops

∑

bandwidth generated · min hops

(2)

In normalized bbp bandwidth block probability is

weighted against the minimum connection distance.

Our rationale behind this parameter, is that long hop

distance connections occupy more FSs in more links,

when compared to more direct connections. If net-

work performance is examined only against BBP, then

a scheme that bans all long connections would super-

ﬁcially perform better. In such a network, if offered

load is sufﬁciently high, then multiple short connec-

tions could be established in the place of a single long

connection. The BBP performance would seemingly

improve, while in reality, if we apply this scheme to

NSFnet, then all east coast to west coast connections

are disabled.

This metric is an attempt to alleviate the unfair-

ness problem that is present in EONs, when high hop-

distance node pairs struggle to establish connections,

even when the network is superﬁcially operating with

low blocking probability. Utilizing normalized bpp

leads to not only improved network performance, but

also slightly decreases the fairness problem. Ignor-

ing this parameter can lead to the undesired result

of rejecting high hop connections and artiﬁcially im-

proving BBP, since then, more low hop connections

would be established. Comparing schemes against

normalized bpp can give additional insights to net-

work performance, since it takes the effect of connec-

tion hop distance into consideration. Our results in

Section 4 demonstrate that our approach not only im-

proves upon this normalized bpp metric, but the ben-

eﬁts can be even observed in the BBP metric (Figures

5 and 6).

Algorithm 2: Calculate masks.

Input : topology: contains the network

topological information

simulation(masks,topology): will

run a simulation and returns the

normalized

bbp

Masks[conn min hops]: limit FS

according to the mask

max f s: maximum FS index

1 masks[i] = [1, max f s]  neutral masks

2 norm bbp = simulation(masks,topology)

3 while True do

4 foreach mask ∈ Masks do

5 mask[2]–  decrease mask

6 new norm bbp =

simulation(mask,topology)

7 if new norm bbp > norm bbp then

8 continue  no improvement

9 if the foreach run once without

improvement then

10 break

Output: The Masks[i] used in Alg. 1

The algorithm that calculates the mask is detailed

in Algorithm 2. In order to design masks that mini-

mize normalized bbp, ﬁrst a baseline is calculated by

using neutral masks for all hop-distances. The neu-

tral mask is in the form of mask[i] = [1, max f sindex],

where i is all possible minimum hop distances for

the network’s nodes. Running a simulation for each

DCNET 2019 - 10th International Conference on Data Communication Networking

topology using those neutral masks produces a base-

line normalized bbp. Then, while the simulations

show that this will improve normalized bbp, each

hop distance’s mask is constantly decreased, until we

have solution. The masks used for each topology in

Section 4 are calculated using this method.

4 PERFORMANCE EVALUATION

4.1 Network Topologies Employed

In order to assess the performance of our approach a

simulation environment was implemented to test our

proposed algorithm HW versus the FF algorithm. In

order to demonstrate the beneﬁts of our proposed al-

gorithm in both elastic backbone and next-generation

metropolitan networks, three topologies were exam-

ined. The backbone network we included in our sim-

ulation is NSFNET (Shen and Tucker, 2009), which

consists of 14 nodes and 21 links (Figure 1). A mesh

based network (Antoniades et al., 2004), which con-

sists of 29 nodes and 41 links (Figure 2), and a ring

topology of 16 nodes are also examined, as represen-

tatives of elastic optical metropolitan networks.

As will be demonstrated in the rest of this Section,

our proposed method achieves different gains depend-

ing on the underlying topology. The beneﬁts in the

case of NSFnet and ring topology are in the region of

10 − 30% in low offered loads, while in the case of

the metro-mesh topology, the highest gain achieved

is 5.5%. Compared to the FF scheme, our method

never underperforms. As our results show, the worst

case performance of HW is at least on par with the FF

method.

4.2 Simulation Parameters

Following the standard settings used in most of the

corpus of EONs, the main characteristics of our sim-

ulation are the following:

• Each link in the network consists of two unidirec-

tional ﬁbers. Each ﬁber contains a maximum of

160 Frequency Slots (FS). This value is typically

used in previous works, and is irrelevant to con-

clusion we derive from our results.

• The source and destination nodes of a trafﬁc re-

quest are uniformly selected.

• The trafﬁc load is dynamically generated and cal-

culated as λ / µ (Erlang). The interarrival connec-

tion rate λ is constant and equal to 1. The con-

nection duration parameter µ follows a negative

exponential distribution.

Figure 5: Bandwidth blocking probability for the NSF net-

work topology. The proposed method can even improve

upon the BBP metric.

• In our implementation of the FF scheme, the pa-

rameter k = 3 for the K-shortest path algorithm.

• The connection load follows one of the two rules:

elastic and 4-7-12. In the case of elastic, the con-

nection load is in the range of [2, 15] FSs, includ-

ing guard-band, while the rule 4-7-12 describes

the scenario of three distinct services with re-

quirements of 4 FSs, 7 FSs and 12 FSs, includ-

ing guardbands (Qiu et al., 2016). In both cases,

the load of newly generated connections is cho-

sen uniformly from the allowed values. Similar

results are observed for both trafﬁc proﬁles. The

average connection load for the elastic trafﬁc is

8.5, while in the case of 4-7-12 is 7.666. This

range of elastic is purposely selected so that the

result lines of both elastic and 4-7-12 in Figures 5-

8 are meaningfully near but not overlapping with

each other.

• The simulation ends when simulation time

reaches 10e5 time units.

4.3 Simulation Results

Even though our method primarily focuses on im-

proving the normalized BBP metric, the improved

network operations can even be of beneﬁt to the more

commonly used BBP metric, as can be seen in Figure

5, leading to gains up to 5.5%. When performance

is compared against the normalized BBP parameter

(Figure 6), the gains are more pronounced reaching

up to 36% for low offer load, though gains dimin-

ish with high loads (Table 1). The topology used

in both of those experiments is the NSFnet (Figure

1). The mask employed for this topology, only lim-

its the 1-hop distance connections to 149 FSs. This

group of experiments demonstrates that our method

A Novel Hop-distance Sensitive Approach to Elastic Optical Networks RSA Algorithms

Table 1: Numerical results comparison for FF and HW for NSFnet.

Erlangs 150 200 250 300 350 400 450 500

FF elastic normalized bbp 0.105 0.215 0.309 0.39 0.455 0.507 0.55 0.587

HW elastic normalized bbp 0.0832 0.192 0.295 0.376 0.441 0.497 0.541 0.577

gains percentage 26.5 11.4 4.9 4 3 1.95 1.81 1.81

FF 4-7-12 normalized bbp 0.08 0.181 0.273 0.352 0.419 0.47 0.51 0.55

HW 4-7-12 normalized bbp 0.058 0.155 0.257 0.336 0.405 0.461 0.50 0.54

gains percentage 36.4 16.79 6.2 4.82 3.58 2.03 1.97 1.34

Figure 6: Normalized bandwidth blocking probability for

the NSF network topology.

will improve the performance of a typical backbone

network, when compared to the FF algorithm. Simi-

lar improvements are observed in both the elastic and

4-7-12 trafﬁc proﬁles.

The NSFnet topology is also used in the results

displayed in Figures 3 and 4. Those Figures give a

detailed image on BBP per N-hop connection for the

former and on the amount of successful/blocked con-

nection per N-hop connection for the latter. Those

results demonstrate both the unfairness that high-hop

connections face and the improvements HW achieves.

Those improvements are more pronounced in com-

parably “low” offered loads, which are also the most

interesting, since a real world scenario will operate

within those ranges of low BBP (a network with 50%

BBP is unresponsive and unrealistic).

Similar results are achieved in both the cases of

the mesh based network (Figure 2) and the ring topol-

ogy of 15 nodes. In both of those topologies, the

mask algorithm limits bandwidth for the 1-hop and

2-hop connections. In the case of the mesh network

(Figure 7), 1-hop connections are limited to 113 FSs

and 2-hop connection are limited to 149 FSs, while

larger hop connections are allow to use the whole

bandwidth. The maximum hop-distance is 6. In the

case of the ring network (Figure 8), 1-hop connections

are limited to 137 FSs and 2-hop connections to 125

FSs. In this topology, the maximum hop-distance is

Figure 7: Normalized bandwidth blocking probability for

the metro network topology.

Figure 8: Normalized bandwidth blocking probability for

the ring network topology.

8. Our experiments show that the ring topology ben-

eﬁts more than the mesh topology when utilizing the

proposed algorithm.

5 CONCLUSIONS

In this work we propose the utilization of the new

metric bandwidth blocking probability normalized to

connection hop distance, or normalized BBP. Our

analysis and the new, weighed probability metric shed

DCNET 2019 - 10th International Conference on Data Communication Networking

light on the intrinsic problem that multihop connec-

tions face in elastic optical networks. High hop con-

nections are harder to satisfy the continuity and con-

tiguity constraints of elastic optical networks, leading

to unfairness and performance deterioration.

Our proposed algorithm achieves up to 10% gains

in BBP when compared to the FirstFit scheme, by

limiting low-hop connections in certain paths and by

preemptively blocking non-shortest-hop connections

when the path utilization is high. Our low complex-

ity algorithm leads to performance improvements that

are shown to be without negative trade-offs.

ACKNOWLEDGEMENTS

This research has been coﬁnanced by the Euro-

pean Union and Greek national funds through the

Operational Program Competitiveness, Entrepreneur-

ship and Innovation, under the call RESEARCH-

CREATE-INNOVATE (project code:T1EDK-05061).

REFERENCES

Antoniades, N., Roudas, I., Ellinas, G., and Amin, J. (2004).

Transport metropolitan optical networking: evolving

trends in the architecture design and computer model-

ing. Journal of lightwave technology, 22(11):2653.

Beletsioti, G. A., Papadimitriou, G. I., and Nicopolitidis,

P. (2016). Energy-aware algorithms for ip over wdm

optical networks. Journal of Lightwave Technology,

34(11):2856–2866.

Beletsioti, G. A., Papadimitriou, G. I., Nicopolitidis, P.,

and Miliou, A. N. (2018). Earthquake tolerant en-

ergy aware algorithms: A new approach to the design

of wdm backbone networks. IEEE Transactions on

Green Communications and Networking, 2(4):1164–

1173.

Chatterjee, B. C., Ba, S., and Oki, E. (2018). Fragmenta-

tion problems and management approaches in elastic

optical networks: a survey. IEEE Communications

Surveys & Tutorials, 20(1):183–210.

Chatterjee, B. C. and Oki, E. (2016). Dispersion-adaptive

ﬁrst–last ﬁt spectrum allocation scheme for elastic

optical networks. IEEE Communications Letters,

20(4):696–699.

Cisco (2016). Vni complete forecast highlights.

Gerstel, O., Jinno, M., Lord, A., and Yoo, S. B. (2012).

Elastic optical networking: A new dawn for the

optical layer? IEEE Communications Magazine,

50(2):s12–s20.

Kyriakopoulos, C. A., Papadimitriou, G. I., and Nicopoli-

tidis, P. (2018a). Exploiting the signal overlap tech-

nique for energy efﬁciency in elastic optical networks.

In 2018 International Conference on Computer, Infor-

mation and Telecommunication Systems (CITS), pages

1–5. IEEE.

Kyriakopoulos, C. A., Papadimitriou, G. I., and Nicopoli-

tidis, P. (2018b). Towards energy efﬁciency in

virtual topology design of elastic optical networks.

International Journal of Communication Systems,

31(13):e3727.

Ning, J. and Shen, G. (2012). Routing and spectrum alloca-

tion in ﬂexi-grid optical networks. In The 2012 11th

International Conference on Optical Communications

and Networks (ICOCN), pages 1–4. IEEE.

Qiu, Y., Fan, Z., and Chan, C.-K. (2016). Efﬁcient routing

and spectrum assignment in elastic optical networks

with time scheduled trafﬁc. Optical Fiber Technology,

30:116–124.

Rosa, A., Wiatr, P., Cavdar, C., Carvalho, S., Costa, J., and

Wosinska, L. (2015). Statistical analysis of blocking

probability and fragmentation based on markov mod-

eling of elastic spectrum allocation on ﬁber link. Op-

tics Communications, 354:362–373.

Shen, G. and Tucker, R. S. (2009). Energy-minimized de-

sign for ip over wdm networks. Journal of Optical

Communications and Networking, 1(1):176–186.

Simmons, J. M. (2014). Optical network design and plan-

ning. Springer.

Vizca

ıno, J. L., Ye, Y., and Monroy, I. T. (2012). Energy ef-

ﬁciency analysis for ﬂexible-grid ofdm-based optical

networks. Computer Networks, 56(10):2400–2419.

Zheng, W., Jin, Y., Sun, W., Guo, W., and Hu, W. (2010).

On the spectrum-efﬁciency of bandwidth-variable op-

tical ofdm transport networks. In Optical Fiber Com-

munication Conference, page OWR5. Optical Society

of America.

A Novel Hop-distance Sensitive Approach to Elastic Optical Networks RSA Algorithms