ArticlePublisher preview available

Dynamic frequency slot allocation on IP-over-EON access links with multiple-type and time-varying traffic

Optica Publishing Group
Journal of Optical Communications and Networking
Authors:
Junyi Shao
Junyi Shao
Junyi Shao
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Shuai Zhang
Shuai Zhang
Shuai Zhang
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Weiqiang Sun
Weiqiang Sun
Weiqiang Sun
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Weisheng Hu at Shanghai Jiao Tong University
Weisheng Hu
Read publisher preview
Download citation
To read the full-text of this research, you can request a copy directly from the authors.

Abstract and Figures

Access links that connect client networks to public networks have to carry multiple-type and time-varying traffic. Those hybrid traffic flows compete for limited bandwidth, yet at the same time have drastically different performance requirements. As planning the link capacity for peak traffic demand is not economically viable, it is important that bandwidth on the access links is shared between different traffic classes in a way that maximal network utility can be achieved. In this paper, we study the frequency slot allocation problem on IP-over-elastic optical network (EON) access links that carry three types of traffic, namely, packet streams, latency critical circuit connections (e.g., video conferencing), and delay tolerant circuit connections (e.g., bulk data transfers). We define four network operation states; in each of which an access link serves traffic with different levels of fulfillment. We then formulate the allocation problem into a weak-constrained optimization problem and propose a genetic algorithm to solve it in real time. Numerical results show that the relative error of the genetic algorithm is within 3% and the access link keeps maintaining the optimal achievable network operation state. We also show that, by increasing the storage size, the access link can adapt to the increasing traffic load within a certain range without upgrading the expensive access link bandwidth. Our study provides useful insights in managing and operating IP-over-EON access links, and the concept of multiple network operation states can be generalized to networks that serve more than one type of traffic.
Figures - available from: Journal of Optical Communications and Networking
This content is subject to copyright. Terms and conditions apply.
A preview of this full-text is provided by Optica Publishing Group.
Content available from Journal of Optical Communications and Networking 
This content is subject to copyright. Terms and conditions apply.
304 Vol. 15, No. 6 / June 2023 / Journal of Optical Communications and Networking Research Article
Dynamic frequency slot allocation on IP-over-EON
access links with multiple-type and
time-varying traffic
Junyi Shao, Shuai Zhang, Weiqiang Sun,* AND Weisheng Hu
State Key Laboratory of Advanced Optical Communication Systems and Networks, Shanghai Jiao Tong University, Shanghai, China
*sunwq@sjtu.edu.cn
Received 17 November 2022; revised 6 April 2023; accepted 15 April 2023; published 2 May 2023
Access links that connect client networks to public networks have to carry multiple-type and time-varying traffic.
Those hybrid traffic flows compete for limited bandwidth, yet at the same time have drastically different per-
formance requirements. As planning the link capacity for peak traffic demand is not economically viable, it is
important that bandwidth on the access links is shared between different traffic classes in a way that maximal
network utility can be achieved. In this paper, we study the frequency slot allocation problem on IP-over-elastic
optical network (EON) access links that carry three types of traffic, namely, packet streams, latency critical cir-
cuit connections (e.g., video conferencing), and delay tolerant circuit connections (e.g., bulk data transfers). We
define four network operation states; in each of which an access link serves traffic with different levels of fulfill-
ment. We then formulate the allocation problem into a weak-constrained optimization problem and propose a
genetic algorithm to solve it in real time. Numerical results show that the relative error of the genetic algorithm is
within 3% and the access link keeps maintaining the optimal achievable network operation state. We also show
that, by increasing the storage size, the access link can adapt to the increasing traffic load within a certain range
without upgrading the expensive access link bandwidth. Our study provides useful insights in managing and
operating IP-over-EON access links, and the concept of multiple network operation states can be generalized to
networks that serve more than one type of traffic. © 2023 Optica Publishing Group
https://doi.org/10.1364/JOCN.481355
1. INTRODUCTION
Access links that connect data centers (DCs) or large institu-
tions to backbone networks can be expensive and are often
the first bottleneck in the entire network. With more remote
access demands caused by working-from-home or learning-
from-home amid the ongoing COVID-19 crisis, efficient use
of access link bandwidth becomes even more important [1–3].
For instance, in an institution with a locally hosted private
cloud service, interactive applications can easily become con-
gested by large file transfers across the enterprise access link if
the link is not carefully managed.
The inter-DC traffic transmitting through access links is
time-varying and unevenly distributed in time, often exhibit-
ing tidal patterns [4,5]. At the same time, network traffic
is composed of different types of flows, each with its own
quality-of-service (QoS) requirements and unevenly dis-
tributed in terms of the number of flows and data volume
in bytes [6–8]. We can roughly divide the traffic into packet
streams and circuit connections. Packet streams such as web
browsing and message transmission are large in number and
require low delay performance, while circuit connections far
exceed packet streams in data volume and prefer low blocking
probability. Circuit connections can be further classified into
latency-critical (LC) flows and delay-tolerant (DT) flows,
depending on the type of traffic they carry [9,10]. LC flows
such as remote video conferences and telemedicine assistance
applications emphasize the timeliness of transmission. In con-
trast, DT flows, e.g., DC backups and large file transfers, have
no requirement on timeliness, and store-and-forward (SnF)
technology can be applied to smooth bandwidth usage during
busy and idle times.
The problem of carrying mixed traffic in backbone networks
that utilize hybrid switching has been widely investigated.
Three types of network architectures that can support hybrid
switching are discussed in [11]. In one of our previous works,
we proposed the BLOC framework for backbone networks to
characterize the dynamic resource allocation problem in hybrid
switching systems [6]. Abundant interest has been given to
the network wide routing, modulation format, and spectrum
assignment (RMSA) problem [12,13]. However, until now,
little attention has been given to the dynamic resource alloca-
tion problem on access links. There exist three challenges. First,
the system performance is essentially multi-dimensional and
involves several QoS requirements corresponding to different
1943-0620/23/060304-14 Journal © 2023 Optica Publishing Group
Storage-assisted optical upstream transport scheme for task offloading in multi-access edge computing
Article
Full-text available
Multi-access edge computing (MEC) applications are often implemented in the form of task offloading, which results in an unprecedented demand for data transfers among MEC servers. However, the combination of expensive and limited bandwidth, growing peak demand, and heterogeneous requirements of mixed traffic has posed a great challenge in terms of task offloading. In this study, we present a storage-assisted optical upstream transport scheme (SOUT) to overcome this challenge. Latency-critical (LC) tasks are given preemptive priority over delay-tolerant (DT) tasks. To reduce peak demand, the storage of an MEC server is introduced to temporarily store DT tasks. Resource partitioning is performed with an adjustable boundary based on traffic fluctuation. Analytic models are presented to investigate the interplay between SOUT and the performance of tasks. Our key findings reveal that there exist two trade-offs to be considered in SOUT. To balance the trade-offs, we formulate the spectrum partitioning and storage assignment problem as an optimization model and solve it using a heuristic approach. Studies show that SOUT provides lower blocking probability for both LC and DT tasks at the cost of slight preemption and limited storage usage when compared with two state-of-the-art optical transport schemes. We further show that 60% of network expenditures can be saved by trading cost-efficient storage for expensive link spectrum resources under a certain network scenario. Overall, this study aims to provide useful insights into task offloading over elastic optical links.
View
Future Optical Networks in a 10 Year Time Frame
Conference Paper
Full-text available
  • Jan 2021
We describe the evolution of the network over a time-scale of 10 years. We cover the merger of photonics and electronics, optical switching, reliance on data-center technologies and architectural evolution.
View
Dimensioning access link capacity for time-varying traffic with mixed packet streams and circuit connections
Article
Full-text available
Access links that connect cloud data centers or enterprise networks to public networks are important parts of the network infrastructure and make up considerable portions of overall network expenditures. When a customer network is connected to an overlaid elastic optical network, determining the capacity of the access link is difficult. Furthermore, access links carry mixed packet streams and circuit connections, each with different performance requirements. Additionally, traffic loads are time-varying and often exhibit diurnal patterns. In this study, we formulate the access link capacity dimensioning problem as an optimization problem with a nested dynamic constraint satisfaction subproblem. Instead of using hard performance measures, such as packet delays and blocking probabilities, we use the probability of fulfillment with regard to different traffic arrival rates as the performance measure and the performance–cost ratio as our optimization objective. Numerical results suggest that a 7.5% to 19.47% cost savings can be achieved if resources are shared between packet streams and circuit connections. And if resource preemptions are allowed, the delay performance for packet streams can be improved by 12% with more frequent preemptions. Our study provides useful insights into the planning and operation of access links.
View
View
View
Argumentation Frameworks with Strong and Weak Constraints: Semantics and Complexity
Article
  • May 2021
Argumentation Framework (AF) has emerged as a central formalism in formal argumentation. Key aspects of the success and popularity of Dung's framework include its simplicity and expressiveness. Integrity constraints help to express domain knowledge in a compact and natural way, thus keeping easy the modeling task even for problems that otherwise would be hard to encode within an AF. In this paper, after providing an intuitive semantics based on Lukasiewicz's logic for AFs with (strong) constraints, called Constrained AFs (CAFs), we propose Weak constrained AFs (WAFs) that enhance CAFs with weak constraints. Intuitively, these constraints can be used to find ``optimal'' solutions to problems defined through CAFs. We provide a detailed complexity analysis of CAFs and WAFs, showing that strong constraints do not increase the expressive power of AFs in most cases, while weak constraints systematically increase the expressive power of CAFs under several well-known argumentation semantics.
View
View
View
View
On the Dynamics of Valley Times and its Application to Bulk-Transfer Scheduling
Article
  • Dec 2020
  • COMPUT COMMUN
Periods of low load have been used for the scheduling of non-interactive tasks since the early stages of computing. Nowadays, the scheduling of bulk transfers-i.e., large-volume transfers without precise timing, such as database distribution, resources replication or backups-stands out among such tasks, given its direct effect on both the performance and billing of networks. Through visual inspection of traffic-demand curves of diverse points of presence (PoP), either a network, link, Internet service provider or Internet exchange point, it becomes apparent that low-use periods of bandwidth demands occur at early morning, showing a noticeable convex shape. Such observation led us to study and model the time when such demands reach their minimum, on what we have named valley time of a PoP, as an approximation to the ideal moment to carry out bulk transfers. After studying and modeling single-PoP scenarios both temporally and spatially seeking homogeneity in the phenomenon, as well as its extension to multi-PoP scenarios or paths-a meta-PoP constructed as the aggregation of several single PoPs-, we propose a final predictor system for the valley time. This tool works as an oracle for scheduling bulk transfers, with different versions according to time scales and the desired trade-off between precision and complexity. The evaluation of the system, named VTP, has proven its usefulness with errors below an hour on estimating the occurrence of valley times, as well as errors around 10% in terms of bandwidth between the prediction and actual valley traffic.
View