You are on page 1of 9

IET Communications

Review Article

Recent trends in MPLS networks: ISSN 1751-8628


Received on 15th November 2018
Revised 6th September 2019
technologies, applications and challenges Accepted on 28th October 2019
E-First on 13th December 2019
doi: 10.1049/iet-com.2018.6129
www.ietdl.org

Mohammad Azmi Ridwan1 , Nurul Asyikin Mohamed Radzi1,2, Wan Siti Halimatul Munirah Wan Ahmad1,
Fairuz Abdullah1,2, Md.Zaini Jamaludin1,2, Mohd Nasim Zakaria3
1Instituteof Informatics and Computing in Energy (IICE), Universiti Tenaga Nasional, 43000 Kajang, Malaysia
2Department of Electrical & Electronics Engineering, College of Engineering, Universiti Tenaga Nasional, 43000 Kajang, Malaysia
3Architecture and Governance, TNB ICT, Kuala Lumpur, Malaysia

E-mail: m.azmiridwan@gmail.com

Abstract: Multiprotocol label switching (MPLS) networks are packet-based networks that offer considerable advantages,
including improved network utilisation, reduced network latency, and the ability to meet the quality of service and strict level
agreement requirements of any incoming traffic. A vast number of applications are now migrating to packet-based conditions
that cause increased pressure on network providers to change their systems. Innovations and improvements on MPLS are still
on-going to ensure that such networks can cater to the ever-increasing bandwidth demand whenever required. This study
provides a review of MPLS networks and their promising technologies, such as traffic engineering, protection and restoration,
differentiated services, and MPLS-transport profile (MPLS-TP) and its applications. This work also reviews recent issues on
MPLS networks and discusses the implementation of MPLS-TP networks in the power grid. A review of recent literature shows
that researchers should be careful in proposing new protocols or designs for MPLS to ensure that it achieves the most efficient
and optimal performance. Furthermore, it can be concluded that although MPLS is a promising technology for future networks,
there are challenges to overcome with regards to security and network flexibility, especially as far as migration to MPLS-TP is
concerned.

1 Introduction path, allocate multiple services in the same network, and treat each
traffic based on QoS requirements.
Circuit-based networks can no longer withstand traffic demand due Given the considerable benefits of MPLS networks, research is
to perpetually increasing bandwidth and delay-sensitive being conducted to ensure that high-bandwidth demand can be
applications. Packet-based applications, such as voice over Internet addressed. Several related reviews on MPLS networks [3–7], with
protocol (VoIP), long-term evolution, and on-demand videos, are their advantages, are summarised in Table 1. However, these
becoming increasingly popular. Hence, current communication studies have focused on only one MPLS technology, whereas our
networks must be improved so that they can manage traffic and review paper addresses broad-ranging MPLS technologies,
fulfil their service level agreement (SLA) [1]. However, network including traffic engineering (TE), differentiated services
designers face challenges in optimising network performance to (DiffServ), protection and restoration, and MPLS-transport profile
achieve the highest efficiency at a reduced cost. (MPLS-TP, an MPLS protocol), in addition to its applications and
Multiprotocol label switching (MPLS) can fully optimise recent issues. To the best of our knowledge, the present work is the
network resources and provide quality of service (QoS) treatment first to discuss recent issues and trends among MPLS technologies.
to the traffic, which has become the de-facto standard for core The organisation of this paper is as follows. Section 2 discusses
network infrastructure. MPLS is scalable, connection-oriented, and the architecture, terminology and advantages of an MPLS network.
independent of any packet-forwarding transport technology. MPLS Section 3 elaborates on the MPLS technologies, including TE,
also reduces the Internet protocol (IP) address look-up at each DiffServ, protection and restoration, MPLS-TP and implementation
router and minimises network latency. MPLS improves packet of MPLS-TP in the power grid. Finally, Section 4 is the
forwarding in a network and overcomes disadvantages of IP conclusions of this study.
forwarding [2]. An MPLS network can decide the best forwarding

Table 1 Recent review papers on MPLS network 2 Architecture and advantages of MPLS networks
Authors Year Reviewed area The Internet Engineering Task Force (IETF) introduced MPLS in
Haddaji et al. [3] 2018 addressed technical challenges 1997 to initially thoroughly address MPLS development. The
encountered by service providers in following issues of MPLS networks were identified:
migrating to MPLS-TP networks
Suhaimy et al. [4] 2018 analysed recent MPLS-TP applications (i) Improving scalability for network layer routing using labels to
aggregate forwarding information.
Eugen [5] 2017 reviewed the performance of MPLS DiffServ
(ii) Improving flexibility in delivering routing services using
Kurimoto et al. [6] 2017 reviewed the MPLS software-defined radio-
MPLS labels to identify traffic with QoS and thus provide special
oriented layer-2 on-demand virtual private
treatment.
network services and network function
(iii) Implementing the label swapping paradigm to optimise
visualisation
networks and thus enhance performance.
Adewale et al. [7] 2016 conducted a comparative simulation study of
(iv) Simplifying router integration with cell switching-based
MPLS for latency and packet loss reduction
technology using common addressing, routing and management
over a wide area network
control.

IET Commun., 2020, Vol. 14 Iss. 2, pp. 177-185 177


© The Institution of Engineering and Technology 2019
LDP and RSVP have advantages and disadvantages. LDP is
preferred in terms of the initial configuration and scalability.
However, RSVP is better suited for migrating from a data link
layer to an MPLS network. Precise planning is required to ensure
the best performance of the MPLS network.

2.1 Advantages of the MPLS networks


The MPLS mechanism can tunnel multiple types of traffic through
Fig. 1  General architecture of the MPLS network
the core network. The tunnel is the path where traffic flows in the
MPLS core network. Tunnelling is a powerful tool because only
ingress and egress routers need to know the content of traffic
carried through the tunnel. Details are hidden from routers at the
core. With the use of MPLS tunnelling, traffic can be explicitly
routed by following traffic policies. Tunnels also provide additional
protection against data spoofing, given that packets can only be
injected at the ingress routers.
An MPLS network offers expenditure reduction by allowing
network operators to control only a single network for all service
Fig. 2  Segments in MPLS header: Label, EXP, S, and TTL
types [8]. This feature is important given that emerging
applications offered by local providers are becoming increasingly
Fig. 1 shows the conceptual design of an MPLS network. dense in terms of traffic and bandwidth consumption. Finally, the
Routers A and C are label edge routers (LERs). Depending on encapsulation of the MPLS overhead, which is only 4 bytes per
which LER is the source or destination, the routers are known as MPLS header, is small and will reduce the latency and workload in
ingress LER or egress LER, respectively. The ingress LER assigns the core network [8].
labels to incoming packets and determines which forward
equivalent class (FEC) these packets belong to. Then, ingress LER
decides the corresponding egress LER and computes the best path 3 MPLS technologies
for the packets to route through the MPLS core network based on In this section, MPLS technologies, including MPLS-TE,
the FEC. Alternatively, egress LER removes the labels from the protection and restoration, DiffServ, MPLS-TP and implementation
packets and forward them using the normal IP forwarding of MPLS-TP in the power grid network are discussed.
procedure.
The core network has label switch routers (LSRs) connected as 3.1 MPLS traffic engineering (MPLS-TE)
either a ring or mesh topology. The LSR reads the label assigned
by the ingress LER and then swaps the label with another that will MPLS-TE is implemented in the network to avoid network
determine the next LSR to be forwarded to. This process is congestion and improve QoS. IETF RFC2720, Section 2.0
repeated until the packet arrives at the destination or egress LER. mentions that MPLS-TE aims to minimise packet loss and delay,
The forwarding decision is based on the fixed-format header. The maximise throughput and support the enforcement of SLA [8].
path computed by ingress LER from the source to the destination is Minimising congestion is the primary objective of TE. Congestion
called a label switch path (LSP). LSR does not store any route per typically occurs under two conditions, namely, insufficient network
the IP forwarding scheme, thereby improving the scalability of the resources and inefficient mapping of traffic onto available
network. Different types of traffic can share a single LSP, and resources. Network congestion can be addressed by expanding the
given sufficient network resources, one LSP can accommodate all network capacity or using classical congestion control by limiting
traffic regardless of their type. However, for critical applications data rate, flow control, queuing management, and schedule-based
that require strict delay and bandwidth treatment, the network control [8]. TE is useful in rerouting the traffic to ensure that data
defines TE or traffic policies. This process is further explained in are transported effectively [9].
the next section. In the case of link or node failures in an MPLS network, TE
Fig. 2 illustrates an example of an MPLS header. The header is verifies that the affected traffic can still reach its destination [10].
divided into four important segments. The first segment is the TE will determine that a specific path has the required
label, which has a size of 20 bits and will be indexed into an MPLS characteristics in accordance with QoS. For instance, high-priority
forwarding table. The next segment is experimental (EXP) bits, applications, especially protection systems, are routed via a path
where classes of services (CoS) are specified for each packet. With that is less congested than others to minimise the delay. Without
vast types of applications tunnelling in the network, the EXP congestion, data loss and delay are reduced, and throughput is
segment is crucial for determining the QoS treatment that will be improved. Consequently, enhanced services can be provided to
provided to the traffic. The next segment is called the bottom-of- consumers.
stack (S) bit. This field is used when more than one label is Local providers can take advantage of MPLS-TE to offer
assigned to a packet. Finally, the time-to-live (TTL) segment is guaranteed bandwidth services that allow clients to have a certain
utilised for path tracing, wherein the value continues to decrease amount of bandwidth available when required. However, the
until it reaches the destination. The packet is discarded when the following scenarios need to be highlighted by network providers
TTL value becomes zero. In an MPLS network, IP headers on the when setting up TE.
packets are still intact, but they are ignored by the LSRs. Instead,
LSRs investigate only the incoming label, go through the labelling (i) The traffic must be forwarded along a predefined path (explicit
table in each LSR, and swap the label with a new outgoing label routing) [11].
immediately. (ii) The utilisation of bandwidth resources must be improved [12].
At the forwarding plane, labels are assigned to each LER and (iii) When resource contention occurs, control must be in place
LSR to find the best LSP. The control plane disseminates over the resource [13].
information from the label. Information can be extracted from the
header in an MPLS network through two approaches. The first uses IETF RFC 3031 provides two options for LSP route selection,
a label distribution protocol (LDP), which is specifically developed hop-by-hop and explicit routing [14]. Hop-by-hop routing allows
to distribute labels. The second approach extends the existing each node to choose the next hop for each FEC traffic
protocol, called the resource reservation protocol (RSVP). LDP independently. In explicit routing, the ingress or egress LSR
offers ease of configuration, session maintenance and reliable specifies the entire LSP [8]. Explicit routing can be achieved by
transport. In RSVP, LDP is extended to allow the creation and implementing RSVP in the network. This process enables the LSP
maintenance of LSPs and create associated bandwidth reservation. to establish the MPLS forwarding state along the path defined at
178 IET Commun., 2020, Vol. 14 Iss. 2, pp. 177-185
© The Institution of Engineering and Technology 2019
devices; multiple controllers can be implemented for networks that
are too complex to be handled [18]. The main goal of SDN is to
allow flexibility for network providers and control the flow of data
through the network. This process enables any traffic policy or TE
to be executed with full flexibility rather than have the same
scheme provided by vendors [19].
Traffic in the MPLS network is normally dynamic and can be
congested at any time throughout the network. TE optimisation
over a single traffic matrix may have some limitations, especially
Fig. 3  Example of MPLS network topology for MPLS-TE explanation when many applications are forwarded in the network because a
single traffic matrix can have large measurement errors and
the source. MPLS-TE uses LSP priorities marked at the header to depicting traffic fluctuations is insufficient. Moreover, large-scale
identify, of which LSPs are more important than others. networks are major challenges for network management to provide
Consequently, the network can confiscate resources from low- QoS guarantees and perform optimisation during network
priority LSPs and guarantee that high-priority LSPs will always be management. The integration of MPLS-TE with SDN can fully
transported first. The high-priority LSP is established along the optimise the network.
shortest and least congested path to ensure maximum throughput Guo et al. [20] presented a multiple-traffic matrix approach to
and minimum delay. Finally, when LSPs need to be rerouted due to solve large measurement errors that occur when the single traffic
link failure, high-priority LSPs will have improved chances of matrix is used. Bahnasse et al. [21] presented a solution using SDN
finding alternative paths as quickly as possible. to manage complex large-scale networks. The TE/SDN network
An example will be discussed using the topology shown in Fig. architecture can effectively manage and provide QoS requirements
3 to provide a thorough understanding of the MPLS-TE process. for multiple service traffic. The reliability of the proposed model is
Supposing Client 1 were to transmit to LSR E, two possible paths dependent on its ability to achieve high-quality VoIP and video
can be used, A-B-E and A-C-D-E. From the operator point of view, with acceptable delay for HTTP response pages. A related issue in
path A-B-E is preferred because it has fewer hops. However, if MPLS-TE is the static bandwidth reservation mechanism of RSVP.
path B-E is congested and has a high latency, the traffic will be The RSVP mechanism in the control plane reserves the same
rerouted via path A-C-D-E. This scenario shows that MPLS does bandwidth at each hop along the tunnel and ignores the difference
not simply select the path with the lowest cost, i.e., the smallest of available bandwidth of other links. This issue rapidly results in
number of hops; instead, it chooses the best path for the traffic to bandwidth exhaustion at the congested link even with underutilised
flow based on QoS requirements. links. TE/SDN can solve this problem by providing the non-
uniform bandwidth reservation to improve the load balancing and
resource utilisation of the network further [22]. By allowing
3.1.1 Issues and related works on MPLS-TE: Kumar et al. [2]
dynamic bandwidth reservation, more protocol label switching
proposed a path protection scheme using MPLS-TE for IEEE 30
tunnels can be computed than that under uniform reservation.
bus system communication for smart grids to improve network
An important feature of the MPLS network is that packets are
resiliency, especially for sensitive protection data, such as SCADA,
forwarded by performing label look-up at the labelling table in
data scanning, and the system refreshment rate for power utility
each LSR. Each application will have different EXP bits, and
networks. MPLS-TE is used and further validated via OPNET to
ingress LER will decide on the best path based on the priority
promptly recover from path failure. Their simulation results
level. However, as the bandwidth and complexity increase, a
showed a reroute time of <10 ms with TE. SCADA is crucial,
shortage of MPLS labels will occur. Huang et al. [23] claimed that
delay-sensitive data; a microsecond of delay will cause inevitable
label consumption is expanding rapidly, thereby leading to
network failure. The implementation of TE in the MPLS network
management complexity, increased operational and capital
can ensure that protection applications are also treated according to
expenditure and table look-up latency, and reduced performance
their QoS requirements.
and scalability. Therefore, the author proposed a method of solving
In multilayer networks, MPLS is usually on top of an optical
the label space reduction problem in the MPLS network using a
transport network. However, different providers usually operate
hybrid MPLS Open-Flow network scheme via LSP multiplexing.
each layer, and information exchange is limited, thereby leading to
This goal was achieved using label stacking and TTL bits to
network degradation. An agreement on the type of information
control packet switching between different LSPs. As a result,
shared from both layers is crucial for significantly improving
traffic with different sources and destinations may share the same
network performance. Therefore, [15] proposed dynamic multicast
LSP, thereby reducing the label space reduction problem with
traffic grooming in MPLS over the optical multilayer network. A
transparent topology
data-learning scheme was utilised on the IP/MPLS layer for logical
According to the abovementioned studies, congestion may still
link cost estimation, and a light path fragmentation-based method
occur even when TE is implemented in a network. Therefore,
used on the wavelength division multiplexing network layer to
additional research is needed to improve network traffic
improve resource sharing in the grooming process. Therefore,
optimisation. Having only a hardware-based MPLS network will
network performance can be greatly improved by managing traffic
restrict local providers in complying with fixed traffic policies
in the network. The use of TE to optimise bandwidth in the
provided by vendors. However, SDN is fully programmable and
network is not sufficient when delay is not considered. Different
offers full flexibility; consequently, the researchers can vary traffic
applications in the network may be delay-tolerant, but some
loads and other performance evaluation parameters, such that
applications can have strict delay requirements. However, only a
innovation can be performed easily.
few related works on TE have focused on both bandwidth and
Even with the benefits of separating the control and data planes,
delay requirements. Thus, Soorki and Rostami et al. [16] presented
as implemented in SDN or network virtualisation, one shortcoming
a new bandwidth- and end-to-end delay-constrained routing
still exists; the lack of rapid and reliable implementation prevents
algorithm that uses data of the ingress and egress node pair in the
the network from growing to its desired capabilities. In an effort to
network.
improve this, Mazhin et al. integrated MPLS with network
virtualisation in their work [24]. This architecture can expand
3.1.2 Software-defined network for MPLS-TE: Software- Internet flexibility and pave the way for the development and
defined networks (SDNs) are emerging due to their flexibility and commercialisation of network virtualisation and next-generation
programmability. With the separation of the control plane from the MPLS. Thus, the integration of MPLS with SDN can substantially
data plane, the network can be managed and innovated easily by impact the future of telecommunication industries.
programming. This feature allows local providers to execute
provision and monitoring effectively and improves network agility
[17]. A centralised single controller manages all the forwarding

IET Commun., 2020, Vol. 14 Iss. 2, pp. 177-185 179


© The Institution of Engineering and Technology 2019
at the failed link is rerouted instead of having protection at the head
end of the entire path. The advantage of FRR is that the network
can choose which resources to protect. In the event of failure,
protection can be applied promptly, and traffic is forwarded to the
rerouted path, which is computed and signalled prior to failure.
Another advantage is that switching time can be improved using
this mechanism. Local protection comes in four variants.

(i) Link protection


(ii) Node protection
(iii) One-to-one protection
(iv) Facility protection
Fig. 4  End-to-end protection using the backup LSP
Link protection refers to the ability to protect traffic from being
forwarded to the LSP in the presence of failure along the LSP. To
protect against link failure, a backup path was set up around the
link for one-to-one protection. Link failure is the most common
type of failure in a network. A link might fail when the link itself
has a problem or when the other link-end has node failure, thereby
disconnecting the entire interconnected link. All four local
protection variants have their own advantages and shortcomings.
Fig. 5  FRR protection by using the detour/bypass LSP However, no protection scheme in the literature can cover all four
[8].
3.2 MPLS protection and restoration To ensure rapid protection, a backup path must be ready to
All traffic in the MPLS network must be delivered with zero packet forward traffic as soon as the failure is detected. To achieve this
loss and at low latency [25]. This requirement is due to the feature, all backup paths must be computed and signalled
bandwidth and delay-sensitive content of important applications, beforehand, and the forwarding state must be set up for switchover.
especially protection data. The network must not suffer from any The forwarding state must be placed at the head and tail ends or the
data discrepancy between the source and destination. This issue merge point (MP) of the backup tunnel or point of local repair
creates a need for a protection and restoration mechanism, which is (PLR) to enable the forwarding of traffic into the backup at PLR
crucial in promptly handling any failure [26]. and back to the main LSP at MP [8]. In the MPLS network, LSR
Traffic with strict SLA requirement, such as video and VoIP, forward the packets through label swapping, and rerouting
has stringent tolerance toward reliability and traffic loss. Similar to decisions are mapped into MPLS labels for protection. Thus, traffic
an MPLS network, immediate recovery after a failure is essential, labels that arrive over the backup tunnel will be the same as those
especially for multiservice networks with applications of different over the failed link. For example, Fig. 5 shows that the label of
priority levels. The MPLS fast reroute (FRR) was introduced by data to be transmitted to Router 3 via Routers 2 and 6 will be the
Cao et al. [25] to provide a guarantee for MPLS tunnels in the same. To ensure that traffic arrives at MP (Router 3) with the
event of a failure; this concept is the same as that offered by correct label, the backup tunnel label must be pushed on top of the
synchronous optical networking (SONET) automatic protection protected LSP label at PLR (Router 2), and penultimate hop-
switching (APS). The primary difference between FRR and popping (PHP) must be performed for the backup tunnel label
SONET APS is that FRR can consistently provide a relatively before MP (Section 3).
small recovery time because the recovery decision is made locally. Network scalability will enable easy expansion and reduce the
The efficiency of the network's recovery was dependent on the inconvenience of network providers in delivering a communication
rates of network failure detection and traffic switching to an network to newly developed areas. In terms of local protection, the
alternative path. This dependency shows that rapid failure detection scalability must be addressed. Stronger protection means additional
is a vital component of MPLS protection. configuration efforts that involve intensive labour and manual path
The first step in providing recovery is detecting failure as soon computations. However, many current implementations can
as it occurs. This process can be done either via hardware-based compute and establish protection paths dynamically. The complex
methods, such as using packet-over-SONET/synchronous digital deployment of a network will impact the number of resources that
hierarchy (SONET/SDH), or non-hardware-based techniques, such can be protected in the network. Network providers must,
as the implementation of an algorithm at a high layer in the therefore, decide on the type of protection (1:1 or N:1 path
network [27]. Effective fault detection and fault notification protections). Moreover, network topologies, such as string-of-
implementation are crucial in providing reliable MPLS protection pearls and dual-plane yield, require long protection paths that may
[27]. Protection comes in two forms, end-to-end protection and increase the forwarding state of the network. The benefit of local
FRR. protection comes at a cost. Hence, understanding how the
End-to-end protection or path protection is commonly used in deployment of local protection affects resource consumption on the
network deployment. Fig. 4 shows that LSP protection is achieved router is crucial [8].
using a primary and backup LSPs. The backup LSP takes over the
traffic in case the primary LSP link fails. Upon receiving the RSVP 3.3 DiffServ-aware traffic engineering
error at the client's node, the primary LSP switches the traffic to the Having MPLS-TE alone still does not guarantee QoS because
backup LSP [28]. One shortcoming of this protection scheme is MPLS-TE is not aware of the DiffServ classes [27]. Therefore,
that traffic will continuously transmit over the failed primary LSP DiffServ is introduced to ensure that TE is aware of the types of
until the RSVP error reaches the head end, thereby causing mode applications for each traffic situation and will be treated based on
delay and data loss. Nonetheless, this option is promising because their QoS requirement. DiffServ-aware aims to eliminate the
it can provide accurate information about where the traffic will requirement of having separate physical networks for different
flow following the failure. However, the backup LSP path may not applications. This feature of MPLS, i.e. the ability to share all
be connected to the same primary router (Router 1). This limitation services with different SLAs in one common core, is attractive.
may not provide meaningful protection in the event Router 1 fails SLA requirements include different scheduling, queuing, and drop
and both backup and primary LSPs are compromised. Thus, having behaviour based on application type and bandwidth guarantees that
path diversity is also a vital issue. are on a per-application basis [29].
The next protection option is FRR, which aims to minimise At present, network providers assign different applications to
delay in the event of traffic failure. Fig. 5 illustrates that the traffic various CoS and mark traffic accordingly. Although this solution is

180 IET Commun., 2020, Vol. 14 Iss. 2, pp. 177-185


© The Institution of Engineering and Technology 2019
main requirement of MPLS-TP is to allow the static creation of
LSP and provide the same features and functionality of
SONET/SDH networks, such as performance monitoring, fault
detection and delay measurement. IETF, in cooperation with ITU,
has defined the requirements for MPLS-TP in IEFT RFC 5654 [8],
including

Fig. 6  Similarities and differences between MPLS-TP and previous MPLS (i) The MPLS-TP data plane must be a subset of the MPLS data
networks plane, as defined by the IETF.
(ii) The design of the MPLS-TP should, as far as reasonably
good, it works well only when network resources are sufficient. possible, reuse existing MPLS standards.
During network congestion, the traffic will decrease or may (iii) Mechanism and capabilities must be able to interoperate with
experience delay and violate SLAs. existing IETF MPLS (RFC3031) and IETF Pseudowire Emulation
In the use of MPLS-TE alone, LSPs are set up by choosing Edge-to-Edge (RFC3985) control and data planes where
links with available resources while ensuring that the bandwidth is appropriate.
always ready for traffic consumption [30]. The only limitation is
(iv) MPLS-TP must support traffic-engineered capabilities that
that MPLS-TE operates only on available bandwidth across all
allow deterministic control of the use of network resources.
classes. In contrast, DiffServ-TE ensures that MPLS-TE is aware
of the CoS by allowing resource reservation and delivers QoS (v) MPLS-TP must support bidirectional transport paths with
guarantees to meet strict SLAs. symmetric bandwidth requirements.
However, SLA still cannot be fully guaranteed even after (vi) MPLS-TP must support static provisioning of transport paths
resources are reserved based on the level of applications with via the management plane.
properly marked traffic. Further mechanisms, such as policing and
admission control, have to be integrated to ensure that traffic stays Additional important requirements for Operations
within limits. At present, network providers use overprovisioning Administration and Maintenance (OAM) are acquired in the
to achieve the goal of service guarantees by ensuring that MPLS-TP network, such as congruency between the data path and
bandwidth is available more than needed. However, OAM, to allow in-band OAM and to have consistent OAM
overprovisioning has its own cost and may only work in normal capabilities for layer 2, pseudowires and LSPs. For improved
cases with no guarantee of success in the event of network failure. network resiliency and protection, the 50 ms protection standard
DiffServ-TE works by first determining the class type of traffic. and various path protections, such as 1:1, 1 + 1, and M:N, must be
The basic DiffServ-TE requirement is to be able to separate supported [33]; in addition, ring topologies and linear and meshed
bandwidth reservation for different classes of traffic. This need networks must be efficiently provided. Nonetheless, the most
requires the network to always keep track of how much bandwidth important architecture goals of an MPLS-TP network are
is available for each type of traffic dynamically at any given time compatibility with the existing MPLS architecture [32], availability
on all paths and routers throughout the network. LSPs that are to static and dynamic provision systems, and support of (without
traffic-engineered to guarantee bandwidth from a class type are dependence on) IP addressing. These goals are a continuation of
called DiffServ-TE LSPs. After traffic classification, the path is using IP addressing, but IP routing is not required because an
computed by calculating all known available bandwidth per class MPLS network uses label switching for packet forwarding. The
type for all priority levels. advantage of using MPLS-TP over the use of its predecessors, such
Traffic is mapped to a DiffServ-TE to the correct scheduler as IP/MPLS and GMPLS, is compatibility with existing MPLS
queue in two ways. technology even with added important capabilities, such as OAM
and resiliency. Some features in previous MPLS are discarded,
(i) EXP bits in the MPLS header are set appropriately at the LSP such as equal cost multipath routing, LDP and PHP.
ingress E-LSP. Fig. 6 shows that MPLS-TP is part of MPLS but with a set of
(ii) The scheduling behaviour is encoded in the forwarding state of new functions satisfying specific requirements of transport
the label of the LSP and EXP bits are used to convey the drop networks. The bidirectional feature (congruency) is needed for
preference for the traffic L-LSP. differential protection relay communications. Static provisioning
behaves similar to SDH, which operates through NMS but provides
Once the traffic is mapped to the correct LSP, it will receive the transport-like resilience with 50 ms protection switching [34].
correct DiffServ treatment. DiffServ provides QoS by dividing Meanwhile, in-band OAM is similar to SDH management, which
traffic into a small number of classes and allocating network performs monitoring, fault detection alarm and multilevel
resources on a per-class basis. Eugen[5] conducted a theoretical operations. Nevertheless, resilience and OAM functions are salient
analysis on MPLS DiffServ using the OPNET modeller simulator, features of MPLS-TP. In terms of resiliency, MPLS already has a
and their simulation results confirmed that DiffServ allows high- rich set of protection and restoration mechanisms, such as LSP
priority traffic to arrive at the destination faster than low-priority FRR and path protection. However, the vital focus in the context of
traffic while improving queuing delay. Meanwhile, MPLS-TE transport-like networks is resiliency, which can trigger protection
enables network resource reservation and optimisation. The from OAM and provide optimised protection in any topology,
integration of DiffServ with TE combines the advantages of both, especially in the ring-type topology, which is the most common
wherein the FRR mechanism improves the overall MPLS network. topology in MPLS networks. Meanwhile, the OAM function helps
Thus, traffic-engineered LSPs with per-traffic class granularity and provide crucial functions, such as reducing operational complexity
guaranteed resources for each traffic and strict QoS requirement and cost, by allowing automatic detection, localisation and failure
can be achieved in normal and network failure cases. handling. It also ensures network availability by locating and
dealing with failures even before clients report them and by
helping maintain SLAs in normal and failure operating conditions.
3.4 MPLS-TP
Network providers are forced to revamp their network 3.4.1 Integration of MPLS-TP with SDN: Similar to the MPLS
infrastructure to minimise operating cost. At present, circuit-based network, MPLS-TP is integrated with SDN to improve network
transport network is evolving to packet-based transport due to the flexibility and integrity by offering programmability, which is
flexibility and cost-benefit offered by packet-switching technology. provided by SDN. Related works have proposed a resource pooling
For this purpose, MPLS-TP is currently being developed to form a mechanism to improve network agility [35], implemented a virtual
basis for next-generation packet transport networks [31]. router based on SDN transport with automatic topology discovery
The goal of MPLS-TP is to provide transport functionality in function with MPLS-TP network [36], developed a generic model
MPLS while preserving the existing MPLS architecture [32]. The transport for SDN and MPLS-TP that enables easy operation of

IET Commun., 2020, Vol. 14 Iss. 2, pp. 177-185 181


© The Institution of Engineering and Technology 2019
intra- and inter-domain network service [6], and reviewed a and maintained stable holdover during topology changes. The
backbone network that uses MPLS-TP and reconfigurable optical interworking of CE and MPLS-TP allowed guaranteed
add-drop multiplexers to create a nationwide network with new performance in critical applications even under extreme conditions.
SDN-oriented on-demand VPN functions [37, 38]. Kim et al. [44] stated that linear protection switching solutions
All related works on integrated SDN and MPLS-TP verify that in packet transport networks, as in MPLS-TP and Ethernet, did not
SDN can improve the networks' protection feature and enhance provide dual-node interconnection capability between two recovery
throughput. SDN reduces the communication between operators of domains. This issue further contributed to the delay, especially
different network layers and can cope effectively with the volatile when many connections were affected by a single failure in
nature of traffic. Furthermore, by using cloud computing, users can multiple recovery domains. Therefore, a collective interconnection
use network resources in carrier networks on-demand. With the use message scheme was proposed in [44] to improve the performance
of virtual equipment, such as the virtual router proposed in [36], of linear protection dual-node interconnection in multiple recovery
flexibility and automated operational control are enhanced by domains. The proposed scheme could perform rapid traffic
means of programming rather than with the use of a hardware- recovery and relieve the communication burden with a recovery
based router. process that was within the 50 ms standard.
Path protection switching management was proposed by
3.4.2 Issues and related works on protection and restoration Rodríguez-Pérez et al. [45] to solve the limited network
in MPLS-TP networks: The majority of related works on MPLS- survivability in multilayer transport during a network failure. A
TP involve protection schemes. Some recent works involve design model for a multilayer network was developed to compare
protection involve provisioning [35, 38–40]. Hayashitani et al. [35] the best utilisation and cost-effective protection scheme. This study
proposed to separate the management fault and working links to revealed that having a single protection process in each layer of the
manage network resources efficiently and thus overcome the network was the best and most cost-effective protection scheme.
inefficiency resource management issue in the event of failure. Azizi et al. [46] extended the wrapping and steering scheme from
Iijima et al. [38] claimed that the MPLS preceding MPLS-TP was [47] for enhanced protection. They showed that the subprotection
not designed to transport power utility data, especially when time was below 50 ms with enhanced time characteristic and
protection traffic is involved. A real-time hardware-in-the-loop resource utilisation. Choi [48] showed a significant increase in data
testing was configured to validate the MPLS-TP network, and the buffer hit ratio as the buffer size was incrementally increased by
study confirmed that the MPLS-TP network could indeed meet the implementing the guarantee of services for the local recovery of
requirements of protection traffic. lost traffic in the network.
Kim et al. [39] proposed the separation of service and
management functions to simplify the performance monitoring of 3.4.3 QoS issues in the MPLS-TP network: The QoS
LSP in the network. The proposed provisioning procedure was mechanism and benefits of MPLS-TP transport networks were
simple, capable of reducing provisioning time, and improves thoroughly explained [49, 50]. However, migrating from SDH to
network scalability. Finally, according to Kolhar et al. [40], MPLS-TP networks has its own challenges. Bandwidth and QoS
increased latency in an MPLS network was due to the slow offered by the current service are no longer sufficient. In [3], the
convergence and complex interactions between switches. A challenges faced by service providers in moving from legacy SDH-
generalised topology discovery, operation monitoring and based to MPLS-TP networks were reviewed, and the results
provisioning controller were designed for carrier-grade transport indicated that the integration of optical and packet-based networks
networks. The fault recovery time was significantly improved by indeed offers simplicity, flexibility and scalability and improved
27 ms while maintaining the traffic load lower than that under results in backhaul networks. The use of MPLS-TP as the metro
Open-Flow. access network offered rapid end-to-end service provisioning.
Several recent works on OAM-related protection are available, Moreover, with TE capabilities, performance monitoring and QoS
such as [34, 41–43]. The concept of mobile MPLS-TP with the use requirement could be enhanced. Providing bandwidth-on-demand,
of OAM channels was proposed by Pijanka and Rózański, [34] to as proposed in [51], also improves the network QoS. As studied in
support the mobility of users and optimise the ‘handoff’ procedure [52, 53], MPLS-TP networks could offer improvement in terms of
in a hierarchical network topology. The combination of mobile QoS while reducing signalling and packet delivery cost.
MPLS with additional OAM functionalities combined the Nonetheless, providing the best capacity with minimal delay and
advantages and capabilities of MPLS-TP technology for mobile low CAPEX and OPEX was challenging. With the use of the
users. The proposed mobile MPLS-TP eased the handoff procedure mathematical model in [54], the overall cost could be minimised by
more than the classic mobile MPLS. Sul et al. [41] claimed that trading off the link capacity cost versus the expected network delay
MPLS-TP networks were still complex despite their promising cost. Separating different services into several links in the network
features and on-the-shelf routers were not reprogrammable. Hence, may seem to be a cost-effective solution, but it does not satisfy the
a hybrid SDN and MPLS-TP network were proposed with a desired performance and increases management complexity and
centralised OAM engine. The hybrid network could meet deployment. Therefore, having an acceptance threshold or
protection requirements. Kim et al. [39] verified the performance benchmark is significant as long as MPLS is implemented in a
of MPLS-TP packet transport OAM and protection mechanism. realistic network environment. Recent works in [3, 46, 48–52]
Their work showed that MPLS-TP could achieve a protection suggested that although MPLS-TP itself may be well-equipped
switching delay below the 50 ms benchmark while protecting LSP with network protection schemes, traffic policies and TE, research
tunnels. Blair et al. [42] proposed a point-to-multipoint MPLS was still needed to serve the perpetually increasing number of
network, which retained the features of TE while reducing load traffic generated in the core network.
with QoS guarantee capabilities. The proposed network managed Table 2 summarises the most recent related works on the MPLS
to transfer data for in-band OAM within 50 ms, which was the state-of-the-art technologies.
protection requirement for OAM packet processing.
Differential protection previously ran on TDM-based networks, 3.5 MPLS-TP in the power grid network
evolving into packet-based network and circuit emulation (CE) is
required. The commonly used protocols were TDM over packet Deploying IP/MPLS or MPLS-TP in smart grid networks is a
(SAToP) and service over packet-switched network (CESoPSN), controversial subject. Equipment developers seek to obtain the best
which were in accordance with ITU-T G.8261. However, Bächli et of both areas through specific management tools and simplify the
al. [43] proposed a novel approach to differential protection signal implementation of IP/MPLS into MPLS-TP platforms. MPLS-TP
transmission over packet-switched networks using adaptive timing networks are still immature due to their remarkably late
circuitry due to issues regarding jitter and asymmetrical delay. introduction, thus becoming debate subjects. In the smart grid,
Frequency stability was implemented in switch hardware that deploying MPLS-TP is conducted not from scratch but as part of
performed similarly to SDH according to ITU-T G.8262 and ITU- an evolutionary transformation plan [57]. At present, smart grid
T G.8264. The proposed technology prevented load impairments communication networks are widely based on SDH/SONET. Time-

182 IET Commun., 2020, Vol. 14 Iss. 2, pp. 177-185


© The Institution of Engineering and Technology 2019
sensitive applications are designed to be transported over circuit- route decisions, as currently presented by SDH/SONET. For IP/
based networks. Synchronous data streams often need to be MPLS, network operations depend on control plane
emulated across the packet core. In the smart grid, all communications. In terms of QoS and deterministic behaviour,
telecommunication network controls are performed in a centralised some utility applications use a dedicated resource, such as separate
network management platform with well-adapted processes and fibres for protection and strictly out of the shared network (offered
skills to operate and maintain the SDN/SONET-based network. In by the MPLS-TP to have more definitive LSPs compared with
this context, MPLS-TP seems to be more suitable than IP/MPLS those in IP/MPLS). Similar to SDH, MPLS-TP responds to all
either with or without TE capabilities [58] because MPLS-TP existing and newly developed packet-based service requirements in
networks have more similarities to SDH/SONET circuit-based utility-sized networks. Traffic rules can be adopted to nodes to
networks. build forwarding labels at any moment. IP/MPLS provides
MPLS-TP forwarding labels are produced by a central network considerable technical capabilities but comes with increased
management system, which allows end-to-end main and backup complexity. Performing MPLS-TE requires good knowledge of
traffic shapes and characteristics, such as in SDH/SONET. Finally,
the transition from SDH to MPLS-TP is feasible due to the similar
Table 2 State-of-the-art technologies for MPLS networks behaviour and network management offered by MPLS-TP. IP/
MPLS Author Year Advantages MPLS is suitable for other types of network operation, such as
technology telecom providers with large and dynamic networks.
integration of Bahnasse et al. 2018 further improved the load Roth et al. [59] presented a case study to look into the
MPLS-TE [21] balancing and resource hybridisation of legacy TDM with MPLS-TP network. The
utilisation of the network traditional TDM network is kept running via TDM services while
with SDN Huang et al. [23] 2018 overcame the label space new packet-switched applications were directly connected to the
reduction problem in MPLS MPLS-TP infrastructure by stages. The proposed hybrid migration
network plan for a utility company was implemented in a power utility in
Guo et al. [20] 2017 improved measurement error Oman's electric grid network for proof of concept. The results
issues when using the single showed a smooth migration path and stated that MPLS-TP fulfilled
traffic matrix the requirement in utility communication network. Additionally,
Mazhin et al. [24] 2017 improved network flexibility Ramirez et al. [60] presented a theoretical-practical evaluation of
MPLS-TP on teleprotection. The authors conducted a laboratory
MPLS-TE Premkumar and 2017 enhanced network resiliency
test using two pieces of teleprotection equipment connected
Saminadan. [2] for sensitive protection data
directly via fibre optic with C37.94 interfaces, and the transmission
MPLS Eugen [5] 2017 presented simulation results time was compared with that of the same equipment connected via
DiffServ that proved the benefits of the MPLS-TP network. The total delay of teleprotection services
DiffServ in terms of delay through MPLS-TP communication was 7.5 ms, whereas that of the
integration of Kurimoto et al. 2017 reconfigurable optical add- direct connection was 4 ms. Nonetheless, the delay for the MPLS-
MPLS-TP [6] drop multiplexers to create TP network remained lower than the latency requirement of 10 ms
nationwide burden for teleprotection. From the analysis performed in the laboratory
with SDN Kurimoto et al. 2016 intra- and inter-domain test, the time response was suitable for energy transport network as
[37] network-enabled generic an optimised path of the progressive migration from the legacy
model transport for SDN and TDM platform to the modern and reliable packet-switched
MPLS-TP network network.
Lim et al. [36] 2016 automatic topology discovery
in the MPLS-TP network 4 Conclusion
MPLS-TP Kim et al. [44] 2018 achieved fast traffic recovery
This study reviews important MPLS technologies, including TE,
protection and relieved the
protection and restorations, DiffServ-aware TE and MPLS-TP. The
communication burden
technologies are explained, and recent issues and related works
Kolhar et al. [40] 2016 reduced fault recovery time were discussed. As transport networks have become increasingly
by 27 ms and decreased congested and complex, a need to migrate from circuit-based to
traffic load carrier-grade packet-based networks has emerged. MPLS is a
Blair et al. [42] 2016 achieved 50 ms protection promising solution for the increasing number of applications that
switching time require different QoS treatments that share the same core network.
Bächli et al. [43] 2016 achieved guaranteed MPLS offers one of the best technologies for dynamically
performance for critical managing traffic with different SLA requirements and overcoming
application even under failure promptly to ensure that consumers can enjoy uninterrupted
extreme conditions services provided by their network providers. This work also
Choi [48] 2016 significant growth of data discusses the expansion of IP/MPLS, which is MPLS-TP. In such a
buffer hit ratio network, important features from the previous MPLS network are
Kim et al. [55] 2016 network availability subject to maintained, whereas inefficient features are removed, and
the maximised subprotection improved protection features are implemented. Aside from telecom
switching time of 50 ms networks, power utilities require a reliable communication system,
Pijanka and 2016 proposed mobile MPLS-TP especially with the migration from legacy grids to smart grid
Rózański. [34] concept reduced the handoff networks, which require the implementation of additional
speed compared with classic applications. This study reviews the implementation of MPLS-TP
mobile MPLS for smart grid networks. Findings show that such implementation is
preferred due to such network's similarities to SDH/SONET-based
MPLS-TP Haddaji et al. [3] 2018 laid out the challenges faced
networks and its suitability for static networks, as in smart grids.
QoS by service providers to
The benefits of MPLS-TP indicate the future of smart grid
migrate from legacy SDH to
telecommunication networks. Although MPLS networks can
MPLS-TP
perform well with sophisticated TE and multi-service aware, it still
Sakamoto et al. 2016 proved that the dynamic needs constant improvement. Based on current issues and research
[56] policing control will efficiently trends, the most common issue is protection, which is the primary
share the bandwidth for concern of every network. Research on protection is on-going to
different services ensure that networks can promptly react to failures. SDN is also

IET Commun., 2020, Vol. 14 Iss. 2, pp. 177-185 183


© The Institution of Engineering and Technology 2019
starting to merge with MPLS-TP due to the flexibility and [27] Virk, A.P.S., Boutaba, R.: ‘Economical protection in MPLS networks’,
Comput. Commun., 2006, 29, (3), pp. 402–408
programmability offered by SDN. Together with the advantages of [28] Calle, E., Marzo, J.L., Urra, A.: ‘Protection performance components in
MPLS, MPLS/SDN can be a promising solution for the ever- MPLS networks’, Comput. Commun., 2004, 27, (12), pp. 1220–1228
increasing congestion and complexity of next-generation networks. [29] Chen, H., Su, H.K., Cheng, B.C.: ‘An objective-oriented service model for
VoIP overlay networks over diffServ/MPLS networks’, Comput. Commun.,
2007, 30, (16), pp. 3055–3062
5 Acknowledgments [30] He, X., Chu, Q., Zhu, M.: ‘Minimizing preemption cost for path selection in
Diffserv-ware MPLS networks’, Comput. Commun., 2006, 29, (18), pp. 3825–
The authors gratefully acknowledge the support and funding of the 3832
TNB Seed Fund (grant no. U-TI-RD-18-07). [31] Kiwon, K., Kwon, T., Kim, S.: ‘Multi-layer resource and path management in
MPLS-TP based packet and optical switched multi-layer transport network’.
Int. Conf. on Information and Communication Technology Convergence
6 References (ICTC), Jeju, Republic of Korea, 2015, pp. 874–879
[32] Hsu, W.-C., Ho, P.-F., Chen, J.-C: ‘Design and implementation of SDN-based
[1] Mushtaq, A., Patterh, M.S.: ‘QoS parameter comparison of diffServ-aware packet transport networks’. Int. Conf. on Wireless Networks, Las Vegas, NV,
MPLS network’. 2017 Int. Conf. on Recent Innovations in Signal Processing USA, 2017, pp. 17–23
and Embedded Systems (RISE), Bhopal, India, 2017, pp. 113–118 [33] Ra, Y., Bang, J., Ryoo, J.-d.: ‘Implementation of FPGA-based MPLS-TP
[2] PremKumar, S., Saminadan, V.: ‘Performance evaluation of smart grid linear protection switching for 4000+ tunnels in packet transport network for
communication network using MPLS’. 2017 Int. Conf. on Communication optical carrier ethernet’, IET Commun., 2018, 13, (5), pp. 481–488
and Signal Processing (ICCSP), Chennai, India, 2017, pp. 2116–2120 [34] Pijanka, M.M., Rózański, G.W.: ‘Mobile MPLS-TP-support the mobility of
[3] Haddaji, N., Nguyen, K., Cheriet, M.: ‘Towards end-to-end integrated optical terminal devices using OAM channel’, Int. J. Electron. Telecommun., 2016,
packet network: empirical analysis’, Opt. Switch. Netw., 2018, 27, pp. 18–39 62, (3), pp. 215–223
[4] Suhaimy, N., Ahmad, W.S.H.M.W., Radzi, N.A.M., et al.: ‘Analysis of [35] Hayashitani, M., Hasegawa, Y., Suzuki, K., et al.: ‘Flexible and automated
MPLS-TP network for different applications’, Int. J. Eng. Technol., 2018, 7, operational control in SDN transport-base virtual router’. Conf. on Optical
pp. 892–898 Fiber Communication, Technical Digest Series, San Francisco, CA, USA,
[5] Eugen, P.: ‘A MPLS simulation for use in design networking for multi site 2014, pp. 5–7
businesses’, vol XVII(1), Economic Sciences Series (Ovidius University [36] Lim, C., Pahk, S., Kim, Y.: ‘Model of Transport SDN and MPLS-TP for T-
Annals, Romania, 2017), pp. 373–378 SDN controller’. IEEE Int. Conf. on Advanced Communications Technology,
[6] Kurimoto, T., Urushidani, S., Yamada, H., et al.: ‘2017 SINET5: a low- Pyeongchang, Republic of Korea, 2016, pp. 522–526
latency and high-bandwidth backbone network for SDN/NFV era’. SDN-NFV [37] Kurimoto, T., Urushidani, S., Yamada, H., et al.: ‘A fully meshed backbone
Track in SAC Symp., Paris, France, 2017 network for data-intensive sciences and SDN services’. Int. Conf. on
[7] Adewale, A.A., Adagunodo, R.E., John, N.S., et al.: ‘A comparative Ubiquitous and Future Networks (ICUFN), Vienna, Austria, August 2016,
simulation study of IP, MPLS, MPLS-TE for latency and packet loss vol. 2016, pp. 909–911
reduction over a WAN’, Int. J. Netw. Commun., 2016, 6, (1), pp. 1–7 [38] Iijima, T., Suzuki, T., Sakamoto, K.: ‘Applying a resource-pooling mechanism
[8] Minei, I., Lucek, J.: ‘MPLS-enabled applications: emerging development and to MPLS-TP networks to achieve service agility’. Cloud Computing 2014:
new technologies’ (John Wiley & Sons Ltd, UK, 2011) The Fifth Int. Conf. on Cloud Computing GRIDs, and Virtualization
[9] Shi, T.J., Mohan, G.: ‘An efficient traffic engineering approach based on flow Applying, Venice, Italy, vol. 15, 2014, pp. 31–36
distribution and splitting in MPLS networks’, Comput. Commun., 2006, 29, [39] Kim, D.U., Kim, B., Lee, J.: ‘A protection switching management of two-
(9), pp. 1284–1291 layer transport networks with MPLS-TP over OTN’. Int. Conf. on ICT
[10] Din, N.M., Hakimie, H., Fisal, N.: ‘Bandwidth sharing scheme in DiffServ- Convergence 2015: Innovations Toward the IoT, 5G, and Smart Media Era
aware MPLS networks’. 2007 IEEE Int. Conf. on Telecommunications and (ICTC 2015), Jeju, Republic of Korea, 2015, pp. 280–284
Malaysia Int. Conf. on Communications, Penang, Malaysia, 2007, pp. 782– [40] Kolhar, M., Abualhaj, M.M., Rizwan, F.: ‘QoS design consideration for
787 enterprise and provider's network at ingress and egress router for VoIP
[11] El-Alfy, E.S., Mujahid, M.S.N., Selim, S.Z.: ‘A pareto-based hybrid protocols’, Int. J. Electr. Comput. Eng., 2016, 6, (1), pp. 235–241
multiobjective evolutionary approach for constrained multipath traffic [41] Sul, D.-m., Byun, S.-h., Kim, T.-i., et al.: ‘A return path bandwidth
engineering optimization in MPLS/GMPLS networks’, J. Netw. Comput. management method for point to multipoint in MPLS-TP’. Int. Conf. on
Appl., 2013, 36, (4), pp. 1196–1207 Information and Communication Technology Convergence (ICTC), Busan,
[12] Awduche, D.O., Jabbari, B.: ‘Internet traffic engineering using multi-protocol Republic of Korea, 2014, pp. 794–797
label switching (MPLS)’, Comput. Netw., 2002, 40, (1), pp. 111–129 [42] Blair, S.M., Booth, C.D., Michielsen, J., et al.: ‘Application of MPLS-TP for
[13] Srivastava, S., van de Liefvoort, A., Medhi, D.: ‘Traffic engineering of MPLS transporting power system protection data’. 2016 IEEE Int. Conf. on Smart
backbone networks in the presence of heterogeneous streams’, Comput. Netw., Grid Communications (SmartGridComm 2016), Sydney, Australia, 2016, pp.
2009, 53, (15), pp. 2688–2702 619–624
[14] Viswanathan, A., Callon, R., Rosen, E.C.: ‘Multiprotocol Label Switching [43] Bächli, R., Kranich, M., Häusler, M., et al.: ‘Teleprotection ensuring highest
Architecture’, 2001, Available at: https://tools.ietf.org/html/rfc3031 performance of the protection system using packet switched wide area
[15] Yu, X., Xiao, G., Cheng, T.H.: ‘An efficient mechanism for dynamic multicast networks (D2/B5 )’. 2016 CIGRÉ Canada Conf., Vancouver, Canada, 2016,
traffic grooming in overlay IP/MPLS over WDM networks’, Opt. Fiber pp. 1–8
Technol., 2014, 20, (4), pp. 341–352 [44] Kim, D.U., Ryoo, J.D., Cheung, T., et al.: ‘Enhanced linear protection
[16] Soorki, M.N., Rostami, H.: ‘Label switched protocol routing with guaranteed switching methods supporting dual node interconnection in packet transport
bandwidth and end to end path delay in MPLS networks’, J. Netw. Comput. networks’, J. Lightwave Technol., 2018, 36, (13), pp. 2638–2650
Appl., 2014, 42, pp. 21–38 [45] Rodríguez-Pérez, F.-J., González-Sánchez, J.-L., Cortés-Polo, D., et al.: ‘A
[17] Arlimatti, S., Hassan, S., Habbal, A., et al.: ‘Software defined network and delay-oriented prioritization policy based on cooperative lossless buffering in
openflow: a critical review’, ARPN J. Eng. Appl. Sci., 2015, 10, (3), pp. PTN domains’, J. Netw. Syst. Manage., 2015, 23, (4), pp. 1016–1033
1244–1252 [46] Azizi, M., Benaini, R., Mamoun, M.B.: ‘Delay measurement in openflow-
[18] Neghabi, A., Navimipour, N., Hosseinzadeh, M., et al.: ‘Load balancing enabled MPLS-TP network’, Mod. Appl. Sci., 2015, 9, (3), pp. 90–101
mechanisms in the software defined networks: a systematic and [47] Francisco-Javier, R., Jose-Luis, G., David, C., et al.: ‘An OAM function to
comprehensive review of the literature’, IEEE Access, 2018, 6, pp. 14159– improve the packet loss in MPLS-TP domains for prioritized QoS-aware
14178 services’, Int. J. Commun. Syst., 2010, 23, (5), pp. 633–652
[19] Kaur, K., Garg, S., Aujla, G.S., et al.: ‘Edge computing in the industrial [48] Choi, J.S.: ‘Design and implementation of a stateful PCE-based unified
internet of things environment: software-defined- networks-based edge-cloud control and management framework for carrier-grade MPLS-TP networks’, J.
interplay’, IEEE Commun. Mag., 2018, 56, (2), pp. 44–51 Lightwave Technol., 2016, 34, (3), pp. 836–844
[20] Guo, Y., Wang, Z., Yin, X., et al.: ‘Traffic engineering in hybrid SDN [49] Addeo, C., Cazzaniga, G., Crescentini, R., et al.: ‘On QoS mechanism
networks with multiple traffic matrices’, Comput. Netw., 2017, 126, pp. 187– profiling in MPLS-TP transport networks’, Bell Labs Tech. J., 2014, 18, (4),
199 pp. 3–17
[21] Bahnasse, A., Louhab, F.E., Ait Oulahyane, H., et al.: ‘Novel SDN [50] Negi, V., Kumar, U., Pandey, T., et al.: ‘Building an MPLS-TP simulator’.
architecture for smart MPLS traffic engineering-DiffServ aware 2012 17th European Conf. on Networks and Optical Communications,
management’, Future Gener. Comput. Syst., 2018, 87, pp. 115–126 Vilanova I La Geltru, Spain, 2012
[22] Huang, L., Shen, Q., Zhou, F., et al.: ‘Nonuniform bandwidth reservation for [51] Orawiwattanakul, T., Otsuki, H., Shimojo, S., et al.: ‘Authorization-based
tunnels in MPLS network using meter tables of openFlow’, Trans. Emerg. approach for customized automated resource provisioning services’. 10th EAI
Telecommun. Technol., 2018, 29, (5), pp. 1–22 Int. Conf. on Testbeds and Research Infrastructures for the Development of
[23] Huang, L., Shen, Q., Zhou, F., et al.: ‘Label space reduction based on LSP Networks and Communities (TRIDENTCOM 2015), Vancouver, Canada,
multiplexing in MPLS openflow hybrid network’, Comput. Commun., 2018, 2015, pp. 1–6
116, pp. 21–34 [52] Cortés-Polo, D., González-Sánchez, J.L., Carmona-Murillo, J., et al.:
[24] Mazhin, G.A., Bag-Mohammadi, M., Ghasemi, M., et al.: ‘Multi-layer ‘Proposal and analysis of integrated PTN architecture in the mobile backhaul
architecture for realization of network virtualization using MPLS technology’, to improve the QoS of HetNets’, EURASIP J. Wirel. Commun. Netw., 2015,
ICT Express, 2017, 3, (1), pp. 43–47 116, (1)
[25] Cao, C., Rouskas, G.N., Wang, J., et al.: ‘Hybrid FRR/p-cycle design for link [53] Garcia, R.C., Daza, B.S.R., Salcedo, O.J.: ‘Evaluation of quality service voice
and node protection in MPLS networks’, AEU - Int. J. Electron. Commun., over internet protocol in WiMAX networks based on IP/MPLS environment’.
2013, 67, (6), pp. 470–478 Proc. 11th ACM Symp. on QoS and Security for Wireless and Mobile
[26] Alouneh, S., Agarwal, A., En-Nouaary, A.: ‘A novel path protection scheme Networks, Cancun, Mexico, 2015, pp. 59–66
for MPLS networks using multi-path routing’, Comput. Netw., 2009, 53, (9), [54] Rassaki, A., Nel, A.: ‘Optimising capacity assignment in multiservice MPLS
pp. 1530–1545 networks’, S. Afr. Comput. J., 2017, 29, pp. 69–89

184 IET Commun., 2020, Vol. 14 Iss. 2, pp. 177-185


© The Institution of Engineering and Technology 2019
[55] Kim, D.-U., Lee, J.Y., Ryoo, J.-d., et al.: ‘Protection switching methods for [58] Cigré: ‘utility communication networks and services’ (Springer International
point-to-multipoint connections in packet transport networks’, ETRI J., 2016, Publishing, Switzerland, 2017)
38, (1), pp. 18–29 [59] Roth, C., Bächli, R., Häusler, M., et al.: ‘How to prepare the communication
[56] Sakamoto, K., Hoshihara, H., Kubo, H., et al.: ‘Bandwidth control using infrastructure for future smart grid applications Oman electricity transmission
adaptive packet policer management for packet transport network based on company SAOC (OETC) sultanate of Oman’. CIGRE Int. Conf. and 22nd
service layer request’, Electron. Commun. Jpn., 2016, 99, (8), pp. 38–47 Exhibition for Electrical Equipment, Muscat, Oman, 2017, pp. 1–9
[57] Samitier, C., Mesbah, M.: ‘Utility communication networks and services: [60] Ramirez, J., Cabrera, H., Bautista, O.: ‘MPLS-TP as packet platform for
specification, deployment and operation’ (Springer International Publishing, critical services in power transmission’. CIGRE Int. Conf. and 21st Exhibition
Switzerland, 2017) for Electrical Equipment, Paris, France, 2016, vol. 6, pp. 1–14

IET Commun., 2020, Vol. 14 Iss. 2, pp. 177-185 185


© The Institution of Engineering and Technology 2019

You might also like