Efficiency and fairness improvement for elastic optical networks using reinforcement learning-based traffic prediction

Anastasios Valkanis; Anastasios Valkanis; Georgios Papadimitriou; Georgios Papadimitriou; Georgia Beletsioti; Georgia Beletsioti; Emmanouel Varvarigos; Emmanouel Varvarigos; Petros Nicopolitidis; Petros Nicopolitidis

doi:10.1364/JOCN.440590

Journal of Optical Communications and Networking
Vol. 14,
Issue 3,
pp. 25-42
(2022)
•https://doi.org/10.1364/JOCN.440590

Efficiency and fairness improvement for elastic optical networks using reinforcement learning-based traffic prediction

Anastasios Valkanis, Georgios Papadimitriou, Georgia Beletsioti, Emmanouel Varvarigos, and Petros Nicopolitidis

Not Accessible

Your library or personal account may give you access

Get PDF
Email
Share
Get Citation
Copy Citation Text
Anastasios Valkanis, Georgios Papadimitriou, Georgia Beletsioti, Emmanouel Varvarigos, and Petros Nicopolitidis, "Efficiency and fairness improvement for elastic optical networks using reinforcement learning-based traffic prediction," J. Opt. Commun. Netw. 14, 25-42 (2022)

Export Citation
- BibTex
- Endnote (RIS)
- HTML
- Plain Text
Citation alert
Save article

Abstract

The characteristics of traffic in core networks have changed in recent years, with the data flows among network nodes now showing large volumes but also large fluctuations during the day. In such a dynamic environment, the static allocation of network resources by making reservations at the peak rates is wasteful, both in terms of energy consumption and resource efficiency. Periodic reconfiguration of network resources is necessary for the full exploitation of an elastic optical network’s potential. In this paper, we present a resource reconfiguration scheme for an IP over elastic optical core network. The cornerstone of the proposed scheme is a novel traffic prediction mechanism based on reinforcement learning. Using the forecasting mechanism, a heuristic routing and spectrum assignment algorithm was designed to efficiently and fairly allocate network resources periodically. Simulation results show both a reduction of the bandwidth blocking probability and an increase of the fairness index regarding the network resources allocation over corresponding algorithms in the literature.

Full Article | PDF Article

More Like This

Optimal and near-optimal alpha-fair resource allocation algorithms based on traffic demand predictions for optical network planning

Tania Panayiotou and Georgios Ellinas
J. Opt. Commun. Netw. 13(3) 53-68 (2021)

Fragmentation-Aware Routing and Spectrum Allocation Scheme Based on Distribution of Traffic Bandwidth in Elastic Optical Networks

Xin Chen, Juhao Li, Paikun Zhu, Ruizhi Tang, Zhangyuan Chen, and Yongqi He
J. Opt. Commun. Netw. 7(11) 1064-1074 (2015)

Tidal-Traffic-Aware Routing and Spectrum Allocation in Elastic Optical Networks

Boyuan Yan, Yongli Zhao, Xiaosong Yu, Wei Wang, Yu Wu, Ying Wang, and Jie Zhang
J. Opt. Commun. Netw. 10(11) 832-842 (2018)

Previous Article Next Article

Cited By

You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Figures (20)

You do not have subscription access to this journal. Figure files are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Tables (3)

You do not have subscription access to this journal. Article tables are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Equations (30)

You do not have subscription access to this journal. Equations are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

	Inputs: Prediction traffic matrix,
	Mean bw demand per request for each OD
	G(V,E) Physical topology
	Output: Routing table
1.	Calculate $k$ -paths for each OD pair
2.	$R o u t i n g_t a b l e \leftarrow k - p a t h s_t a b l e$
3.	$c o n g e s t e d_l i n k s \leftarrow \emptyset$
4.	$n o t_r e r o u t e d_O D \leftarrow {{o d}_{1, 2}, {o d}_{1, 3}, \dots, {o d}_{i, j}}$
5.	$f l a g \leftarrow f a l s e$
6.	Calculate link congestion
7.	while $n u m_c o n g e s t e d_l i n k > 0$ AND flag = false do
8.	sort (congested_links) according to expected load
9.	for $⟨ e a c h e \in c o n g e s t e d_l i n k s ⟩$ do
10.	$c a n d i d a t e_O D \leftarrow \emptyset$
11.	for $⟨ e a c h {o d}_{i, j} \in n o t_r e r o u t e d_O D ⟩$ do
12.	if $e \in {p a t h}_{i, j}$ then
13.	$c a n d i d a t e_O D \leftarrow c a n d i d a t e_O D \cup {o d}_{i, j}$
14.	end if
15.	end for
16.	for $⟨ e a c h {o d}_{i, j} \in c a n d i d a t e_O D ⟩$ do
17.	${c o s t}_{i, j =} d_{new} - d_{old}$
18.	end for
19.	sort (candidate_OD) according to cost
20.	for $⟨ e a c h {o d}_{i, j} \in c a n d i d a t e_O D ⟩$ do
21.	if feasible path found then
22.	${p a t h}_{i, j} \leftarrow {n e w}_{p a t h}$
23.	$n o t_r e r o u t e d_O D \leftarrow n o t_r e r o u t e d_O D -$
${{o d}_{i, j}}$
24.	break
25.	end if
26.	end for
27.	if feasible path found then
28.	Calculate link congestion
29.	$R o u t i n g_t a b l e \leftarrow {n e w}_{r o u t i n g}_t a b l e$
30.	break
31.	else
32.	$f l a g \leftarrow t r u e$
33.	end if
34.	end for
35.	end while
36.	Return Routing table

	Inputs: Prediction traffic matrix,
	Mean bw demand per request for each OD
	G(V,E) Physical topology
	threshold, M modulation formats
	Routing table
	Outputs: G’(V,E) Updated physical topology
	congested links, num_congested_links
1.	for $⟨ e a c h {o d}_{i, j} \in O D ⟩$ do
2.	for $⟨ e a c h a \in {p a t h}_{i, j} ⟩$ do
3.	$L_{l} \leftarrow L_{l} + [c e i l ({m r}_{ij} / F S * M_{ij}) + G B] * {n r}_{ij}$
4.	end for
5.	end for
6.	for $⟨ e a c h e \in E ⟩$ do
7.	if $L_{e} / C ⟩$ congestion_threshold
8.	$c o n g e s t e d_l i n k s \leftarrow c o n g e s t e d_l i n k s \cup e$
9.	$n u m_c o n g e s t e d_l i n k s \leftarrow n u m_c o n g e s t e d_l i n k s + 1$
10.	$E \leftarrow E - e$
11.	end if
12.	end for
13.	Return G’ congested links, num_congested_links

	Inputs: Prediction traffic matrix,
	Mean bw demand per request for each OD
	G(V,E) Physical topology
	$R_{r e s}$ , M modulation formats
	Routing table
	Outputs: Reserved channels for OD pairs
	FIRST STAGE
1.	$L o n g_d i s t \leftarrow \emptyset$ ; $S h o r t_d i s t \leftarrow \emptyset$ ; $h \leftarrow c e i l (h_{max} / 2)$
2.	for $⟨ e a c h {o d}_{i, j} \in O D ⟩$ do
3.	if ${D i s t}_{i j} > h$ then
4.	$L o n g_d i s t \leftarrow L o n g_d i s t \cup {o d}_{i, j}$
5.	else
6.	$S h o r t_d i s t \leftarrow S h o r t_d i s t \cup {o d}_{i, j}$
7.	end for
8.	for $⟨ e a c h {o d}_{i, j} \in L o n g_d i s t ⟩$ do
9.	create clusters of od not having common links
10.	end for
11.	$N c \leftarrow f l o o r$ (W/num_clusters)
	SECOND STAGE
1.	for $⟨ e a c h {o d}_{i, j} \in L o n g_d i s t ⟩$ do
2.	calculate expected $L_{i, j}$ for ${o d}_{i, j}$
3.	if ${o d}_{i, j}$ not rerouted OR ( ${o d}_{i, j}$ rerouted AND
	${o d}_{i, j}$ has no common links with other od of its
	cluster)
4.	${R e s_F S}_{i, j} \leftarrow m i n$ (floor ( $L_{i, j} \times R_{r e s}$ ), Nc)
6.	end if
7.	end for
8.	Return Reserved spectrum channels for distant OD pairs

	Inputs: Prediction traffic matrix,
	Mean bw demand per request for each OD
	G(V,E) Physical topology
	Output: Routing table
1.	Calculate $k$ -paths for each OD pair
2.	$R o u t i n g_t a b l e \leftarrow k - p a t h s_t a b l e$
3.	$c o n g e s t e d_l i n k s \leftarrow \emptyset$
4.	$n o t_r e r o u t e d_O D \leftarrow {{o d}_{1, 2}, {o d}_{1, 3}, \dots, {o d}_{i, j}}$
5.	$f l a g \leftarrow f a l s e$
6.	Calculate link congestion
7.	while $n u m_c o n g e s t e d_l i n k > 0$ AND flag = false do
8.	sort (congested_links) according to expected load
9.	for $⟨ e a c h e \in c o n g e s t e d_l i n k s ⟩$ do
10.	$c a n d i d a t e_O D \leftarrow \emptyset$
11.	for $⟨ e a c h {o d}_{i, j} \in n o t_r e r o u t e d_O D ⟩$ do
12.	if $e \in {p a t h}_{i, j}$ then
13.	$c a n d i d a t e_O D \leftarrow c a n d i d a t e_O D \cup {o d}_{i, j}$
14.	end if
15.	end for
16.	for $⟨ e a c h {o d}_{i, j} \in c a n d i d a t e_O D ⟩$ do
17.	${c o s t}_{i, j =} d_{new} - d_{old}$
18.	end for
19.	sort (candidate_OD) according to cost
20.	for $⟨ e a c h {o d}_{i, j} \in c a n d i d a t e_O D ⟩$ do
21.	if feasible path found then
22.	${p a t h}_{i, j} \leftarrow {n e w}_{p a t h}$
23.	$n o t_r e r o u t e d_O D \leftarrow n o t_r e r o u t e d_O D -$
${{o d}_{i, j}}$
24.	break
25.	end if
26.	end for
27.	if feasible path found then
28.	Calculate link congestion
29.	$R o u t i n g_t a b l e \leftarrow {n e w}_{r o u t i n g}_t a b l e$
30.	break
31.	else
32.	$f l a g \leftarrow t r u e$
33.	end if
34.	end for
35.	end while
36.	Return Routing table

	Inputs: Prediction traffic matrix,
	Mean bw demand per request for each OD
	G(V,E) Physical topology
	threshold, M modulation formats
	Routing table
	Outputs: G’(V,E) Updated physical topology
	congested links, num_congested_links
1.	for $⟨ e a c h {o d}_{i, j} \in O D ⟩$ do
2.	for $⟨ e a c h a \in {p a t h}_{i, j} ⟩$ do
3.	$L_{l} \leftarrow L_{l} + [c e i l ({m r}_{ij} / F S * M_{ij}) + G B] * {n r}_{ij}$
4.	end for
5.	end for
6.	for $⟨ e a c h e \in E ⟩$ do
7.	if $L_{e} / C ⟩$ congestion_threshold
8.	$c o n g e s t e d_l i n k s \leftarrow c o n g e s t e d_l i n k s \cup e$
9.	$n u m_c o n g e s t e d_l i n k s \leftarrow n u m_c o n g e s t e d_l i n k s + 1$
10.	$E \leftarrow E - e$
11.	end if
12.	end for
13.	Return G’ congested links, num_congested_links

Abstract

Cited By

Figures (20)

Tables (3)

Equations (30)

Journal of Optical Communications and Networking