Content uploaded by Afeez Oke
Author content
All content in this area was uploaded by Afeez Oke on Aug 20, 2022
Content may be subject to copyright.
1. Introduction
In the last few decades, technologies for the planning,
development and operation of communication networks
have largely remained unchanged. With the exponential
growth of the modern Internet, network specications
and network trac have changed signicantly, as
rigorous technologies are continually evolving for cloud
services, IoT (Internet of Things) technologies and the
new generation of mobile communication system of 5G,
server virtualization and big data (Hakiri et al., 2014; Li
et al., 2020; Singh & Jha, 2017). The growing complexity
of traditional networking architectures does not cater for
the exponential increase of data trac (Rehman et al.,
2019). In addition, technology progresses, connectivity
needs and the rise of numerous resources, such as
VOIP or streaming video in High Denition where
unimaginable when conventional network architecture
was planned. These contributed to the issue of improving
the network infrastructure control and adapting it to
the recurring resource needs of modern applications.
Cognitive networks and software dened networks
are evolving quickly to satisfy the current networking
demands of isolation, virtualization, trac engineering,
Technoscience Journal for Community Development in Africa 2:1 (2021) 69–79
ISSN 2659-0573
Research Article
A scheme for resilient routing in residue number system based
software dened networks
Oke Afeez Adeshina
*,1
, Akinbowale Nathaniel Babatunde
2
, Oloyede Abdulkarim Ayopo
3
, Kazeem Alagbe
Gbolagade
2
1 Department of Computer Science, College of Natural and Applied Sciences, Summit University, Oa, Nigeria
2 Department of Computer Science, Faculty of Communication and Information Technology, Kwara State University, Malete, Nigeria
3 Department of Telecommunication, Faculty of Communication and Information Sciences, University of Ilorin, Ilorin, Nigeria
Abstract: In recent times, the advances in internet technology and the rise of network trac, owing to relentless creation of
rigorous and mission-critical applications, has been giving serious attention to resilient routing in Software Dened Network
(SDN). The evolution has made it possible for extensive use of SDN to respond to the current requirements of modern
networking. Residue Number System (RNS) been used to minimize latency in routing tables for Openow protocol. However,
majority of network elements remain vulnerable to failures, especially transmission devices and links. It has been shown that
proactive approaches to link recovery in RNS based SDN can select alternate path during link failures, however, the challenges
of fast failure recovery still exist with large data path labels for both primary and alternate routes. Furthermore, the problem
of early backup path failure before primary path still remains a challenge. The paper presents an approach that is intended to
improve SDN routing and reduces the computational in RNS based SDN. The proposed scheme utilizes the shortest path re-route
algorithm and a reactive mechanism to respond to link failure when it occurs. The proposed scheme which will be implemented
as a prototype using the mininet emulator and oodlight controller is intended to eectively route packets especially when there
are link failures for RNS based SDN.
Keywords: Residue Number System (RNS); Software Dened Network (SDN); resilience; OpenFlow; trac engineering; controller
* Corresponding author:
Email: okeafeez@summituniversity.edu.ng
Proof!!! Proof!!! Proof!!!
70 Technoscience Journal for Community Development in Africa, Vol 2, No 1, 2021
access control and above all programming (Martinello
et al., 2014). These types of networks are required
since existing networks are outdated due to new and
innovative technologies and the limit to network growth
is being made available (Malik & Fréin, 2020).
Figure 1: Traditional network layer versus SDN
The data plane and control plane are merged in
traditional network architectures, as shown in gure
1, the interconnection of the Traditional layer makes it
closed as it layer depends on each other. Additionally, the
control plane does not hold any abstraction levels unlike
the data plane. (Kreutz et al., 2015). As seen in gure 1,
the SDN network layer work independent of each other.
The Control Plane consists of conguration protocols
such as Multi-Protocol Label Switching (MPLS) and a
variety of routing protocols. In Traditional Networks,
example of protocols include, Routing Information
Protocol (RIP), Enhanced Interior Gateway Routing
Protocol (EIGRP), Shortest Path Bridging (SPB) etc.
However, this protocol addresses particular problems
without basic abstraction. This causes network diculty
in connecting a computer. In addition due to packet
ooding, increased time to track errors, estimating
alternative paths and upgrading the router table, there
is a signicant recovery delay in conventional IP
networks (Alzahrani & Fotiou, 2020). This makes
existing networks very stagnant and cannot be regulated
dynamically.
The next generation infrastructure for the internet
has implemented the SDN concept to tackle traditional
network architecture issues in order to create an optimal
architecture that can be managed eectively, adaptably
and at low costs (Braun & Menth, 2014; Hakiri et al.,
2014; Zhao et al., 2019). The SDN architecture has
established a network abstraction layer. As shown in
gure 2, the data and control planes are characterized
by network services abstraction (Saraswat et al., 2019).
A centralized authorization known as the controller in
gure 2, provides the communication to instruct network
switches to route and monitor the trac through the
network (Braun & Menth, 2014). As the Controller
has a global networking view, it would provide ideal
network routes either before or on request. (Alzahrani &
Fotiou, 2020). With the global view, it thus determines
optimally for the disrupted ows when looking for an
eective alternative route. Additionally, a well-dened
programming interface between the switches and the
SDN controller facilitates the distinction between
control and data plane as depicted in gure 2. (Braun &
Menth, 2014; Kreutz et al., 2015).
Figure 2: SDN architecture
SDN Controller congures transition ows and
tracks how nodes are communicated by the switch
(Braun & Menth, 2014). Therefore, it is possible to
detach network devices on the switch and control trac
on the OSI layer. This is simpler than the traditional
network separation.
OpenFlow is the most common and open standard
communication protocol between the SDN controller and
OpenFlow switches. OpenFlow was designed in 2008
and currently managed by Open Network Foundation at
Stanford University (ONF). The fundamental principle
of OpenFlow is that there are forwarding tables and
an open API that the OpenFlow controller uses. The
transmitting rules to the switch are being used as the
controller knows the Network Vision. The opened
specication requires separate devices from multiple
vendors to be connected to one OpenFlow controller.
(Braun & Menth, 2014).
A scheme for resilient routing in residue number system based software dened networks / O. A. Adeshina et al. 71
While OpenFlow is a signicant step towards
opening the control plane, it does not simplify hardware
entirely or allow ow state changes. (Martinello et al.,
2014). OpenFlow architectures often have scalability
problems in core networks in particular since they
involve a broad number of active ow-related states
(Braun & Menth, 2014).
OpenFlow’s limitations led researchers to construct
smoother core network components for packet forwarding
using the Residue Number Method. (Liberato et al.,
2018; Martinello et al., 2014, 2017; Spalla et al., 2015).
By means of a combination of small numbers produced
by the remaining part of a large number, RNS makes it
possible to represent a large number entirely. The large
number is a route ID, the small numbers are the output
ports of the core switches used for the specic path and
the related co-primes are the core network interfaces.
Current networks are very dynamic, and therefore, the
conguration changes and link failures in a single time
are very high (Rehman et al., 2019; Yu et al., 2019).
Consequently, many resilient routing approaches in
SDN networks based on RNS routing were proposed in
this regard. (Gomes et al., 2016; Liberato et al., 2018;
Spalla et al., 2015), however, the conguration of
backup paths to achieve high network throughput and
latency in failure recovery still remains a challenge.
With substitution of table lookups by modulo
operations, RNS based SDN schemes were able to
mitigate latency problems occurring in OpenFlow
Lookup table in SDN (Cercós et al., 2014; Martinello
et al., 2014), however, the challenges of fast failure
recovery still exists (Muthumanikandan, 2017).
Additionally, the Residue Arithmetic based schemes
where fault detection mechanism were included are
all proactive based approaches (Liberato et al., 2018;
Valentim et al., 2019). Here, the use of larger data path
labels for both primary and emergency routes and the
problem of early backup path failure before the primary
path are challenges (Valentim et al., 2019).
Network elements, including forwarding devices
and connections, are likely to fail (Malik & Fréin, 2020;
Rehman et al., 2019). As a result, network facilities such
as routing are damaged. Recovery from failure can be
done either proactively, or reactively also referred to as
recovery. The alternate path is prepared and reserved
for security before failure happens. However, the
solution is not pre-planned during restoration and can be
automatically determined (on request) if a malfunction
happens. Link failure in software dened networks are a
continuing source of service interruption. Consequently,
many re-routing methods have been suggested after
a connection breakdown. This study focuses on the
problem of multiple contingency paths for the treatment
of single and dual link failure. In the current study, we
focus on a reactive approach for fast failure recovery
in other to reduce the large data labels and also build
multiple alternate emergency path in case of multiple
link failures for a RNS based SDN. The shortest path
fast re-routing technique is employed in this scheme
in order to reduce the loss of data and minimize the
recovery time.
The next section of the paper provides an overview
of RNS together with the common representations and
denitions. Various related work on RNS based Software
Dened Network are illustrated in Section III. Section
IV discusses resilient routing in Software dened
networks with the proposed scheme and implementation
in section V and VI. Finally, the paper is concluded in
Section VII with a discussion on anticipated results and
the direction for further research.
2. Overview of residue number system
2.1. Overview of Residue Number System (RNS)
The RNS is a system of unusual numbers identied
with the relatively prime moduli set { ,, ...,
} which is gcd( ) = 1 where (Navi et al.,
2011). Considering a number X with a modulo m, then
X could be expressed by its residue r as (1).
r = |X|m (1)
Equation (1) can be represented by X ≡ r (mod m).
For instance, the number X=64 with modulo m =
13 can be shown as . Similarly, the
number Z=51 with the same modulo could be represented
as = = 12. The modulus 13 shows both X and
Z as 12. In this case, X and Z are congruent modulo m.
Instead of utilizing a single modulus, a set of multiple
moduli, { ,, ..., }, is proposed to represent a
number. A number can then be represented as one set
of residues, accordingly.
For example, assuming there is one set of moduli {13,
16, 19}, number X and Z would be represented in
RNS as and
respectively.
The general RNS calculation model as shown in
Figure 3 below includes the conversion steps to the RNS
and the positional notation from the RNS is depicted.
72 Technoscience Journal for Community Development in Africa, Vol 2, No 1, 2021
Additionally, a separate computational structure
is used to represent dierent kinds of non-modular
operations (Deryabin et al., 2018). RNS data transfer
requires that the Chinese Remaining Theorem (CRT)
be used by a forward or opposite converter to transform
weighted operands to residual representations (Navi
et al., 2011) or the Mixed Radix Conversion (MRC)
(Gbolagade & Voicu, 2011; Navi et al., 2011). The
forward conversion includes the transition to RNS
equivalent of binary or decimal numbers, while the
reverse conversion transforms RNS numbers to binary
or decimal numbers, and arithmetical operations like
addition, subtracting and multiplying can be carried
out in parallel by using RNS without carries between
residue numbers.
Chinese Remainder Theorem (CRT)
Let the residue representation of x for moduli
be given as . The
Chinese Remainder Theorem makes it possible to
determine |x|M , provided the greatest common divisor
of any pair of moduli is 1. Using the equation:
(2)
Where and is the multiplicative inverse
of with respect to such that .
Such that
12 3
11 1
11 1 2 2 2 3 3 3
|| | | | | |
mm m
X mM x mM x mM x
−− −
=++
(3)
RNS for routing in Software Dened Networks
Let Network domain represent a set of n Switches in
a desired path so that where
need to be pairwise relatively primes.
That is the switches represent the moduli set.
Let P be a set of outgoing ports
which is considered as a residue.
Thus, in order to establish communication between
two pair of hosts, the RNS based Controller needs to
calculate what is the number (route- ID) for which the
modulo operations results will lead the packets to their
destination.
Then, there exists a unique integer X such that 0 ≤ X <
that solves the congruence.
.
Let
Using CRT it is possible to calculate X through its
residues:
Figure 3: Model of computation in RNS
A scheme for resilient routing in residue number system based software dened networks / O. A. Adeshina et al. 73
(4)
Where and is the multiplicative inverse
of with respect to
(5)
For example, from the diagram in Figure 4, core
switches are assigned to co-prime numbers to establish
the communication between the pair of hosts, in this
example {5, 7, 11, 13}. For instance, it chooses the route
to be set through the switches S = { , , } = {13,
5, 7} and switches’ output ports are P = { , , } =
{2, 0, 0}. A label (route id) is computed using CRT, for
example 210 in the example diagram M =13 · 5 · 7 =
455 and =35, =91, =65.
= = = 3
= = 1
= = 4
Using CRT, R = < · · + · · + ·
· > mod M
R = < 210 +0+0 >mod 455 = 210.
On computing the route-ID, the controller sends the
route-ID to the edge switches in example above 210 which
then install the respective ow-table rule. Additionally,
the ingress edge switch is responsible for embedding
route-ID into each packet coming from src host to dst
host. Thus, as shown in the Figure above, when a packet
enters a core, the modulo of the route ID and the switch is
used to determine the output port to forward the packet to.
For example when S13 receives a packet with route-ID
(R = 210), it forwards the packet to port
; then, S5 forwards it to port ; after, S7
forwards it to port , reaching the egress
edge switch that removes the route-ID from the packet
and delivers it to dst host.
3. Related works
3.1. Residue number system based software dened
networks
RNS was used by Wessing et al., (2002) for packet
forwarding and to avoid both optical header rewriting
of photonic packet switching networks and the need
for label distribution protocols. In packet forwarding in
Figure 4: Sample Packet forwarding using Residue Number system based SDN’s.
74 Technoscience Journal for Community Development in Africa, Vol 2, No 1, 2021
software-dened networks, RNS is often used to replace
the OpenFlow table lookup with modulus of the RNS
operations that allow explicit recognition route coding
at the edges and stateless forwarding techniques in the
center of the network (Liberato et al., 2018; Martinello
et al., 2014). The use of RNS for packet forwarding has
the potential to unravel the eciency of SDN usage in
core networks. The section reviews literatures that have
addressed or use RNS to improve SDN’s.
Wessing, et al. (2002) applied RNS to avoid both
optical header rewriting of photonic packet switching
networks and the need for label distribution protocols.
The authors implemented a packet transmission using
the Chinese Remainder Theorem (CRT) by creating a
label based on two arrays. Both of these arrays have all
the node-specic keys and all the output information
needed for a particular path through the core network.
Where the network path is wanted, an array consists
of all node-specic keys and the other array consists
of the output ports for the specied path. The authors
congured the network with each core node and
assigned a key, however, the information of the topology
of the network and the assigned keys are provided to
the edge nodes. The authors computed the label using
CRT when the packet enters the network and nally
the label is added to the packet that is transmitted to
the core network. Their system is able to calculate the
output information within each of the core nodes using
the modulo operation of RNS. The authors assessed
the system’s scalability, allowing the software to scale
up to 50 core nodes for multi-protocol label switching
(MPLS) networks, since only 7 bytes are needed for the
label in the header. However, the minor restriction in
route length and the requirement for larger number of
nodes in core networks still remains a challenge.
Martinello et al. (2014) proposed a scheme to
further decouple the control and data planes in the
design of SDN. Their approach totally eliminated the
use of lookup tables in the forwarding engine using
the operations of RNS. The route path ID is computed
based on RNS without further interaction with the
nodes. Experimental analysis was performed using
Mininet emulation environment and OpenFlow 1.0
with a Round Trip Time (RTT) above 50 percent and
30 percent reduction in keeping active ow state in the
network. However, the scheme is not fault tolerant as
there are no protection paths that can be pre-computed
at the controller.
Cercós et al. (2014) performed a detailed and
comprehensive data plane power consumption analysis
of the OpenFlow switch by breaking it down its design
modules The KeyFlow was proposed as an alternative
since it eliminates a ow table lookup. Experimental
analysis reveals that the overall power consumption is
reduced by 53.7%.
Cercos et al. (2015) proposed a new south-
bound protocols within SDN with the use of Residue
Number System which the SDN architecture in terms
of cost and energy ecient forwarding engine. This
authors used the Keyow in (Martinello et al., 2014)
approach to simultaneously reduce latency, jitter, and
power consumption in core network nodes of SDN.
Experimental result reveals that the round-trip time
(RTT) can be reduced above 50% compared to the
OpenFlow protocol, especially for densely populated
ow tables. The author’s demonstrated the reduction
of jitter by implementing the prototype on a NetFPGA
based platform, which reveals that there is a 57.3%
reduction in power consumption.
Spalla et al. (2015) proposed a scheme to explore
the OpenFlow roles for the design of resilient SDN
architectures relying on multiple controllers. The authors
implemented their scheme using the Ryu controller
exploring the OpenReplica service to ensure consistent
state among the distributed controllers and a prototype
was tested with the RouterBoards/MikroTik switches
and evaluated for latency in failure recovery and switch
migration for dierent workloads.
Gomes et al. (2016) proposed a scheme incorporating
deection guiding mechanism and RNS in intra-domain
resilient routing system in which edge-nodes set a
route ID to select any existing route as an alternative to
safely forward packets to their destination. The authors
used the RNS approach in (Martinello et al., 2014) but
improved the Network by dealing with failed links
based on driven routing deections that enables to keep
the communication alive even without the controller
reaction to a failure. Experimental analysis using Mininet
network emulation environment shows that the scheme
was able to avoid packet loss, and using the deection
controls the enabled the packet disordering on TCP
throughput to about 25%. However, source routing from
the controller takes long time to recover from failures.
Liberato et al. (2018) proposed a new concept of
programmable Residues Dened Networks (RDN) based
on SDN concept. The authors used a protection mechanism
which permits tableless switches in a faster alternative to
controller-based restoration of the route to re-route packs
directly onto the data plane. For the sake of fault tolerance,
an alternate path is pre-computed into the packets that lead
the packets to their destination if a key switch detects a
fault in their networks. The authors implemented the RDN
prototype in Mininet emulated environment while the
core switches were implemented in NetFPGA devices to
A scheme for resilient routing in residue number system based software dened networks / O. A. Adeshina et al. 75
increase the accuracy on latency measures. Experimental
result shows the scheme oers an ultra-fast failure recovery,
and no jitter within the RDN core while also achieving low
latency in around 2.33μs for 3 hops and 2.93μs for 4 hops
with RDN switching time per hop (≈ 0.6μs). However,
there is overhead due to the insertion of both primary and
emergency route in the packet header, the use of emergency
route can be optimized for multiple link failures.
Valentim et al., (2019) proposed a Residue Dened
Network Architecture for load balancing by exploiting
the elephant ow isolation and strict source routing in
core nodes. The authors used the ow classication
operations performed on the edge using features of the
OpenFlow protocol. The scheme used core switches
that forward packets based on the remainder of division
between the route identier and the switch identier.
A prototype was implemented using OpenStack as the
cloud manager framework. Edge and core switches were
virtualized using a customized version of OpenvSwitch
(OvS). Experimental result show that the scheme is
able to migrate routes with low data loss rate, without
compromising the communication between servers.
However, the scheme is not able to detect congestion
detection and queue overow detection.
Lacan & Lochin, (2020) introduced a XOR-based
source routing enable fast forwarding and low- latency
communications. The scheme utilizes linear encoding to
construct the route labels of unicast data and multicast
data transfers. In contrast to standard table lookup, they
allow quick, computationally ecient routing decisions
without any packet alteration along the path. The authors
compared their scheme with the scheme proposed by
(Liberato, et al., 2018) that uses RNS modular property
to compute a label-ID number to identify (following a
reverse operation) the output switch port considering a
unique router ID. The comparative analysis indicates
that the scheme computes the smallest label possible
and presents strong scalable properties compared to
approaches based on modular arithmetic. However,
their scheme does not possess a fault tolerant capability
and it involves a lot of complexities.
3.2. Resilient routing in software dened networks
Resilient routing and link failure recovery in software
dened networks is becoming more important nowadays
because of increasing development of new internet
and devices, there has been an extraordinary change
in network requirements coupled with an increase
in network trac due to the constant development of
rigorous applications and increasing number of users.
Hence, there is need for ecient and fast failure
recovering during packet routing (Muthumanikandan,
2017; Petale, et al., 2020; Rehman et al., 2019; Yu et al.,
2019). Link failure recovery techniques take advantage
of central control SDN unique features and exibility
of programmable data planes for real-time applications
like video conferencing and voice over IP (VOIP),
which can tolerate a delay of 50 ms in case of recovery.
The dierent methods of link failures recovery in SDN
can be divided widely into two categories: proactive and
reactive (Petale et al., 2020).
3.2.1. Proactive approaches for link recovery
In proactive recovery, alternate backup routes have been
precongured. The identication of failure is local and
the ows from the failed link are passed automatically
without contact with the control unit to the alternate
direction. If a fails, the ow rules for the backup path
are set on the switch already. Therefore the packets from
the failed connection are diverted to the alternate route
identied by the switch. Proactive Recovery proponents
contend that proactive recovery is time ecient, because
paths are preset and no alternate paths must be tested by
the controller. Therefore, it is held to a minimum every
time you consult the controller and select an alternate
course. An alternate path, however, must be built
for each ow of the failed link, which is impractical,
violating the limitations placed on the ow table input
by the data plane switches.
3.2.2. Reactive Approaches for Link Failure Recovery
in SDN
Reactive failure recovery is primarily reliant on the
SDN controller (Lin, et al., 2016; Petale et al., 2020).
If the controller senses faults, then by sending regular
heartbeat messages the controller searches for a possible
route for the failing link. The controller eliminates the
old ow inputs and attaches new ow inputs to the SDN
switches for the revised route.
While the reactive methods the alternate path can
be sought dynamically, control intervention causes
a considerable duration of recovery. Because of the
overall synchronization between the switches and the
control, there is additional time to locate the alternate
path and to initiate new ow entries.
4. The proposed scheme and implementation
The proposed system uses a reactive mechanism to
respond to link failure. The Controller uses RNS
arithmetic for routing instead of the OpenFlow table
routing, when a link failure occurs, the switch nearest
76 Technoscience Journal for Community Development in Africa, Vol 2, No 1, 2021
Figure 5: Proposed qrchitecture of RNS based SDN with shorte path fa rerouting
Proposed Resilient re-routing procedure
Start:
For communication between source host to destination host
Step 1: Start
Step 2: Extract Ethernet header
Step 3: if (Ethernet == type of Source Routing)
Extract SR header & get Port
Compute route id using CRT
Set Output Port and emit packet
If (Output is not available)
Calculate Alternate Path using Shortest re-route
Repeat Step 3
Endif
End if
Step 4: Deliver to Output port
Step 5: Stop
A scheme for resilient routing in residue number system based software dened networks / O. A. Adeshina et al. 77
to the failure sends a signal to the Controller which then
computes the shortest path from the switch to the required
destination. The architecture of the proposed scheme,
showing the details of the RNS path computations and
the shortest path fast re-route is shown in gure 5.
Shortest path fast rerouting
The set of shortest paths as shown in gure 6, are computed
reactively on demand thereby memory overhead can be
reduced. Multiple backup shortest paths are computed to
help in handling single and dual link failures.
Figure 6: The shortest path algorithm
The embedding of resilient and fast link failure
recovery modules in SDN is of crucial importance. Most
proposals for resiliency and fault tolerance in RNS based
SDNs are mostly proactive with large overheads for
emergency route embedding in packet header (Liberato
et al., 2018; Martinello et al., 2017). Others have used
route deection which is probabilistic and has transient
loops (Gomes et al., 2016).
The ep taken for reactive response to link failure
using shorte path re-route is explained below:
- The controller calculates the backup path
reactively upon notication of a link failure and
installs it in the switch.
- The switch generates renewal packets to keep
the existing backup paths alive irrespective of
the idle time of ow entries
- The new route-ID generated is used to perform
modulo operations on the switches to determine
the next switch and the destination
- Once the failed link is up, switch noties the
controller to recalculate the best path by sending
recovery packets.
5. Results and discussion
In cases of link failure, most approaches have used
route restoration which tries to notify the controller
to recalculate the route by excluding the faulty links
from the available path. Other approaches include route
deection (Gomes et al., 2016) which is probabilistic
and has transient loop and the selection of an
Emergency route Id (ERI) (Liberato et al., 2018). Our
approach uses a reactive approach to link failure. For
the implementation of RNS based routing algorithm and
shortest path fast re-route, the RNS tables routing was
improved on by introducing the shortest path fast re-
route during link failure. When a packet arrives at the
ingress edge switch, the packet is sent to the Controller
which selects the main routes among all pre-calculated
paths between the source and destination. Additionally,
the Controller installs OpenFlow rules at the ingress and
egress switches, when a link failure occurs, the nearest
switch to the failure point communicates to the controller
which then reactively compute the set of alternate routes
using the shortest path fast re-route algorithm.
When link failure occurs, a message is sent to
the controller which chooses from the shortest path
computed for re-routing. The controller computes the
protection path based on the model shown in gure 5,
ensuring there are no emergency route necessary thereby
reducing packet header information and increasing
routing resilience especially for multiple link failures.
In situations where there are no link failures, there is no
overhead in computing the alternate path in the packet
header. The controller only responds when there is a link
failure. When the switch receives a packet, it checks if the
port is available, then primary route ID is used. However,
if the switch port is not available, the controller selects
a new primary route ID using the shortest path fast re-
route algorithm, henceforth, the new route-ID is used to
bypass the failed link. For the subsequent switches along
the route, they just keep doing the modulo operation until
the packets reach the edge.
6. Conclusion and further work
This research develops a proposal for RNS based SDN
with the shortest path re-route for resilient routing.
The algorithm would be introduced to SDN with the
possibility of reducing the early backup path failure
and the overhead caused by larger data path for both
78 Technoscience Journal for Community Development in Africa, Vol 2, No 1, 2021
primary and emergency route in Residue arithmetic
based SDN. The next intent is to create a prototype
of this methodology, test and evaluate with the RNS
based proactive approach to resilient routing in SDN.
The expectation here is that the proposed approach
will successfully and eciently create a fast reaction
to failure, reducing the large data path for the primary
and emergency route as well as the early failure of
emergency route in the proactive approach.
References
Ali, J., Lee, G., Roh, B., Ryu, D. K., & Park, G. (2020).
Software-Dened Networking Approaches for Link
Failure Recovery : A Survey.
Alzahrani, B., & Fotiou, N. (2020). Enhancing Internet of
Things Security using Software-Dened Networking.
Journal of Systems Architecture, 110(January), 101779.
https://doi.org/10.1016/j.sysarc.2020.101779
Aremu, I. A., & Gbolagade, K. A. (2017). An overview of
Residue Number System. 6(10), 1618–1623.
Braun, W., & Menth, M. (2014). Software-Dened
Networking Using OpenFlow: Protocols, Applications
and Architectural Design Choices. Future Internet, 6(2),
302–336. https://doi.org/10.3390/6020302
Cercos, S., Ramon, M., Ewald, A. C., Moisés, R. N.,
Manolova, A., Monroy, T., & Salda, S. (2015). networking
Design of a stateless low-latency router architecture
for green software-dened networking. https://doi.
org/10.1117/12.2077560
Cercós, S. S., Oliveira, R. E., Vitoi, R., Martinello, M.,
Ribeiro, M. R. N., Fagertun, A. M., & Monroy, I. T.
(2014). Tackling openow power hog in core networks
with keyow. Electronics Letters, 50(24), 1847–1849.
https://doi.org/10.1049/el.2014.2346
Deryabin, M., Chervyakov, N., & Tchernykh, A. (2018). High
Performance Parallel Computing in Residue Number
System High Performance Parallel Computing in Residue
Number System. 9(February), 62–67.
Gbolagade, K. A., & Cotofana, S. D. (2009). An O(n) residue
number system to mixed radix conversion technique.
Proceedings - IEEE International Symposium on Circuits
and Systems, 1(1), 521–524. https://doi.org/10.1109/
ISCAS.2009.5117800
Gbolagade, K. A., & Voicu, G. R. (2011). An Ecient FPGA
Design of Residue-to-Binary Converter for the Moduli
Set {2n+1,2n,2n-1}. IEEE Transactions on Very Large
Scale Integration (VLSI) Systems, 19(8), 1500–1503.
https://doi.org/doi: 10.1109/TVLSI.2010.2050608
Gomes, R. R., Liberato, A. B., Dominicini, C. K., Ribeiro,
M. R. N., & Martinello, M. (2016). KAR: Key-for-
Any-Route, a Resilient Routing System. Proceedings
- 46th Annual IEEE/IFIP International Conference on
Dependable Systems and Networks, DSN-W 2016, 120–
127. https://doi.org/10.1109/DSN-W.2016.11
Hakiri, A., Gokhale, A., Berthou, P., Schmidt, D. C., &
Gayraud, T. (2014). Software-dened networking:
Challenges and research opportunities for future internet.
Computer Networks, 75(PartA), 453–471. https://doi.
org/10.1016/j.comnet.2014.10.015
Kreutz, D., Ramos, F. M. V., Verissimo, P. E., Rothenberg,
C. E., Azodolmolky, S., & Uhlig, S. (2015). Software-
dened networking: A comprehensive survey.
Proceedings of the IEEE, 103(1), 14–76. https://doi.
org/10.1109/JPROC.2014.2371999
Lacan, J., & Lochin, E. (2020). XOR-based Source Routing.
IEEE International Conference on High Performance
Switching and Routing, HPSR, 2020-May(February).
https://doi.org/10.1109/HPSR48589.2020.9098991
Li, Y., Su, X., Ding, A. Y., Lindgren, A., Liu, X., Prehofer,
C., Riekki, J., Rahmani, R., Tarkoma, S., & Hui, P.
(2020). Enhancing the internet of things with knowledge-
driven software-dened networking technology: Future
perspectives. Sensors (Switzerland), 20(12), 1–20.
https://doi.org/10.3390/s20123459
Liberato, A., Martinello, M., Gomes, R. L., Beldachi, A.
F., Salas, E., Villaca, R., Ribeiro, M. R. N., Kanellos,
G., Nejabati, R., Gorodnik, A., & Simeonidou,
D. (2018). RDNA: Residue-Dened Networking
Architecture Enabling Ultra-Reliable Low-Latency
Datacenters. IEEE Transactions on Network and Service
Management, 15(4), 1473–1487. https://doi.org/10.1109/
TNSM.2018.2876845
Liberato, A., Martinello, M., Gomes, R. L., Beldachi, A. F.,
Salas, E., Villaca, R., Ribeiro, M. R. N., Kondepu, K.,
Kanellos, G., Nejabati, R., Member, S., Gorodnik, A.,
& Simeonidou, D. (2018). RDNA : Residue-Dened
Networking Architecture. 15(4), 1473–1487.
Lin, Y., Teng, H., Hsu, C., Liao, C., & Lai, Y. (2016).
Fast Failover and Switchover for Link Failures and
Congestion in Software Dened Networks.
Malik, A., & Fréin, R. De. (2020). applied sciences Rapid
Restoration Techniques for Software-Dened Networks.
https://doi.org/10.3390/app10103411
Martinello, M., Liberato, A. B., Beldachi, A. F., Kondepu,
K., Gomes, R. L., Villaca, R., Ribeiro, M. R. N., Yan,
Y., Hugues-Salas, E., & Simeonidou, Di. (2017).
Programmable residues dened networks for edge data
centres. 2017 13th International Conference on Network
and Service Management, CNSM 2017, 2018-Janua,
1–9. https://doi.org/10.23919/CNSM.2017.8255987
Martinello, M., Ribeiro, M. R. N., Oliveira, R. E. Z. De, &
Vitoi, D. A. (2014). KeyFlow: A Prototype for Evolving
SDN Toward Core Network Fabrics. April, 12–19.
Molahosseini, A. S. (2012). Research Challenges in Next-
Generation Residue Number System Architectures. Iccse,
1658–1661.
A scheme for resilient routing in residue number system based software dened networks / O. A. Adeshina et al. 79
Muthumanikandan, V. (2017). Link Failure Recovery Using
Shortest Path Fast Rerouting Technique in SDN. Wireless
Personal Communications. https://doi.org/10.1007/
s11277-017-4618-0
Navi, K., Molahosseini, A. S., & Esmaeildoust, M. (2011).
How to Teach Residue Number System to Computer
Scientists and Engineers. In IEEE Transactions on
Education, 54(1), 156–163. https://doi.org/doi: 10.1109/
TE.2010.2048329.
Omondi, A., & Premkumar, B. (2007). Residue Number
Systems: Theory and Implementation. Imperial College
Press.
Petale, S., Member, G. S., & Thangaraj, J. (2020). Link
Failure Recovery Mechanism in Software Dened
Networks. 8716(c), 1–8. https://doi.org/10.1109/
JSAC.2020.2986668
Rehman, A. U., Aguiar, R. L., & Barraca, J. P. (2019). Fault-
tolerance in the scope of Software-Dened Networking
(SDN). IEEE Access, 7, 124474–124490. https://doi.
org/10.1109/ACCESS.2019.2939115
Saraswat, S., Mishra, R., & Gupta, A. (2019). Challenges
and Solutions in Software Dened Networking : A
Survey Challenges and Solutions in Software Dened
Networking : A Survey. May. https://doi.org/10.1016/j.
jnca.2019.04.020
Singh, N. (2016). An overview of Residue Number System.
August.
Singh, S., & Jha, R. K. (2017). A Survey on Software Dened
Networking: Architecture for Next Generation Network.
Journal of Network and Systems Management, 25(2),
321–374. https://doi.org/10.1007/s10922-016-9393-9
Spalla, E. S., Maoletti, D. R., Liberato, A. B., Rothenberg,
C., Camargos, L., & Martinello, M. (2015). Resilient
Strategies to SDN : An Approach Focused on Actively
Replicated Controllers, 246–259. https://doi.org/10.1109/
SBRC.2015.37
Valentim, R. V., Villaca, R. S., Ribeiro, M. R. N., Martinello,
M., Dominicini, C. K., & Maoletti, D. R. (2019). RDNA
Balance: Load Balancing by Isolation of Elephant Flows
using Strict Source Routing. 1–3.
Wessing, H., Christiansen, H., Fjelde, T., & Dittmann, L.
(2002). Novel scheme for packet forwarding without
header modications in optical networks. Journal of
Lightwave Technology, 20(8), 1277–1283. https://doi.
org/10.1109/JLT.2002.800268
Yu, Y., Li, X., Leng, X., Song, L., Bu, K., Chen, Y., Yang,
J., Zhang, L., Cheng, K., & Xiao, X. (2019). Fault
management in software-dened networking: A survey.
IEEE Communications Surveys and Tutorials, 21(1),
349–392. https://doi.org/10.1109/COMST.2018.2868922
Zhao, Y., Li, Y., Zhang, X., Geng, G., Zhang, W., & Sun, Y.
(2019). A Survey of Networking Applications Applying
the Software Dened Networking Concept Based on
Machine Learning. IEEE Access, 7, 95397–95417.
https://doi.org/10.1109/ACCESS.2019.2928564
