On the Performance comparison of RIP, OSPF, IS-IS and EIGRP routing protocols

Vasos Hadjioannou
Department of Computer Science
University of Nicosia
Nicosia, Cyprus
e-mail: [email protected]

Abstract— Nowadays, routing protocols have become a

crucial part of the modern communication networks. A II. DISTANCE VECTOR ROUTING PROTOCOLS
routing protocol's responsibility lies in determining the way
routers communicate with each other in order to forward any A. Routing Information Protocol (RIP)
kind of packets, from a source to a destination, using the RIP is a distance-vector routing protocol that is
optimal path that would provide the most efficiency. There are based on the Bellman-Ford algorithm and uses hops as a
many routing protocols out there today, some old and some routing metric. It avoids loops by limiting the number of
new, but all are used for the same purpose. In general, to
ideally select routes between any two nodes on a computer
hops that are allowed in a single path, from a source to a
network and disseminate information. This paper takes into destination. The maximum number of hops allowed are 15,
consideration four of such routing protocols (RIP, OSPF, IS-IS which limits the size of networks RIP can support. If a route
and EIGRP), expresses them and analyzes their way of contains more than 15 hops, it is then considered as an
operation. It also presents the results of a simulation, that took infinite distance and its destination is unreachable. RIP uses
place for the sole purpose of studying the behavior of those the User Datagram Protocol (UDP) as its transport protocol,
four protocols, under the same circumstances, as well as the and is assigned the reserved port number 520.
evaluation of the comparison with one another based on the
results of the simulation. Bellman-Ford algorithm:
The Bellman–Ford algorithm is an algorithm that
I. INTRODUCTION computes shortest paths from a single source vertex to all of
the other vertices in a weighted graph. It is similar to
When talking about routing protocols, a very Dijkstra, but instead of greedily selecting the minimum-
important question inevitably comes up: "Which protocol to weighted node, not yet processed, to relax, it simply relaxes
use?". The full-proof way to use in order to decide, is to set all edges, and does this |v|-1 times. The repetitions allow
up a network and use the protocols that might possibly suit minimum distances to propagate accurately, since in the
the network's needs. This will prove to be extremely absence of negative cycles, the shortest path can visit each
inefficient and expensive in time, money and resources. To node at most once.
give an answer to this question, network simulations come Bellman-Ford can detect cycles and report their
into play. existence, and its running time is O(|v|*|e|) where v denotes
A network simulation is a technique for modeling the number of vertices in the directed graph, and e the
the behavior of a network by calculating the interaction number of edges [1].
between the different network entities, using mathematical
formulas, or actually capturing and playing back
observations from a production network. Such simulations
can be performed by various tools (such as OPNET, ns,
NetSim etc.) otherwise known as network simulators.
A network simulator is a software that was
designed to predict the behavior of a computer network by
evaluating the performance of a model that corresponds to
the network that needs to be analyzed.
In this case, the OPNET simulator was used to Figure 1: Bellman-Ford algorithm pseudo code
evaluate the performance of the four aforementioned
protocols and compare them with each other, by sending IP RIP operation:
datagrams and traffic from one computer to another, over a Every RIP router maintains a routing table, and for each
modeled network. destination there is one routing entry:
  the shortest distance from the router to the EIGRP is an enhanced distance-vector protocol
destination which relies on the Diffused Update Algorithm (DUAL) to
 next router along the path to the destination calculate the shortest path to a network.

The distance from the router to the destination is EIGRP has four methods[4]:
periodically advertised by the routers to their neighbors.  Neighbor Discovery/Recovery
When a router receives such an advertisement, it checks  Reliable Transport Protocol (RTP)
whether or not any of the advertised routes are useful in  Diffusion Update Algorithm (DUAL)
improving the current routes. If so, the current route of the  Protocol Dependent Modules (PDM)
router is updated, in order for it to go through the
advertising neighbor.
As mentioned before, routes are compared by their III. INTERIOR GATEWAY PROTOCOLS
length, where their value needs to be between 1 and 16 (16 Both of the following Interior Gateway Algorithms
means infinity). In order for loops to be avoided, routers (OSPF and IS-IS) are based on the Dijkstra algorithm.
must do split horizon when they advertise a route to their
neighbor. Dijkstra:
The complete routing table is advertised by a router The Dijkstra algorithm is a graph search algorithm
to all of its neighbors, and for each route there are two that solved the single-source shortest path problem for a
timers ("timeout" and "garbage-collection" time). graph with non-negative edge paths, producing a shortest
A destination will be marked as unreachable when path tree. It maintains a set of vertices (S) whose shortest
a route hasn't been refreshed for three minutes, letting the path from the source has already been determined. It
router assume that there is a link failure. This will set the repeatedly selects the vertex with the minimum shortest
timer of the garbage collection to 2 minutes, and if the timer path, adds it to the set, and then relaxes all edges leaving
expires before the entry is updated, the route will then be from it.
removed from the routing table [2].
RIP disadvantages: Steps for executing the Dijkstra algorithm:
 Maximum hop size is 15 1. Assign to every node distance value from the source
 Variable length subnet masks are not supported by - 0 for the source
RIP version 1 - ∞ for all other nodes
 Slow convergence and count-to-infinity problem 2. Mark all nodes unvisited and set the initial node as
"current". Create a set of unvisited nodes consisting all the
B. Enhanced Interior Gateway Routing Protocol(EIGRP) nodes except the initial one.
3. For the current node, consider all of its unvisited
The Enhanced Interior Gateway Routing Protocol neighbors and calculate their tentative distances. If the new
is an advanced distance-vector routing protocol, designed by distance is less than what already exists, we override the old
Cisco Systems, in ordered to help automate routing distance with the new one.
decisions and configuration on a computer network. EIGRP 4. When we are done considering all the neighbors of the
was designed to support classless IPv4 addresses, and for current node, mark it as "visited" and remove it from the
this reason, it replaced the Interior Gateway Routing "unvisited" set.
Protocol (IGRP), which could not. 5. If the destination node has been marked "visited" ---> exit
Routers use EIGRPs discovery and recovery the search
mechanism in order to dynamically learn of other routers on 6. Otherwise, select the unvisited node with the smallest
their directly attached networks. They must also find out tentative distance and mark it as current. Repeat from step3.
whether or not their neighbors are reachable, or indeed
operational, which is achieved by having each router send
small Hello packets periodically. A router can determine the A. Open Short Path First (OSPF)
operation of its neighbor as long as Hello packets keep OSPF is an Interior Gateway Protocol (IGP), based
getting received by it, and once the availability of its on the Dijkstra algorithm, used for routing Internet Protocol
neighbors is made positive, routers can exchange routing (IP) packets within autonomous systems, or other routing
information [3]. domains. It is a link-state routing protocol. Such protocols
The rules by which traffic is forwarded is kept in a are also referred as SPF (Shortest Path First)-based or
routing table which many routers have, and if a valid path to distributed database protocols [5].
a destination is not known by a router, the traffic is In link-state routing, each node constructs a map,
discarded. Furthermore, routers automatically share in the form of a graph, which contains the connectivity
information, which makes EIGRP a dynamic routing information about the network. This map shows how the
protocol. nodes are connected to one another and which node to
which. Every node will then calculate the best logical path
to every possible destination in the network, leading to the Features of IS-IS [29]:
creation of the node's routing table, by collecting the best  Hierarchical routing
paths individually calculated by each node. After the routing  Classless behavior
table is created, it is presented to the Internet Layer which  Rapid flooding of new information
routes datagrams according to the address located in the IP  Fast convergence
packets [6].  Scalable
OSPF is able to detect link failures, or any other
changes in the topology, and adapt within seconds by Operation of IS-IS:
converging to a new loop-free structure. Shortest Path First Each router running IS-IS sends a hello packet to
algorithms, in general, make for a stable network since they every interface that has IS-IS enabled in order to discover its
require frequent, but small, updates and converge quickly, neighbors and establish adjacencies. If the information
thus avoiding routing loops and the count-to-infinity contained in a hello packet meets the criteria for forming an
problem. On the other hand, they take up a lot of CPU adjacency, then routers that share a common data link will
power and memory. become IS-IS neighbors. After that, routers flood Link-State
OSPF supports Internet Protocol Version 4 (IPv4), Packets (LSP) to their neighbors, except the ones who they
Internet Protocol Version 6 (IPv6), Variable Subnet received the same LSP from. Then, all routers will construct
Masking (VLSM) and Classless Inter-Domain Routing their individual Link-State Database and a shortest-path tree
(CIDR) addressing models. (SPT) is calculated by each IS, which is used to construct
The portable devices (e.g. smart phones, tablets etc.) in the routing table [8].
emerging mobile network architectures have shrunk in size,
incorporating advanced functions and mechanisms as IV. SIMULATION
explored in [7-15]. Authors in [16] use a novel algorithm to The simulation was performed using Opnet IT
handle the traffic with the self-similarity properties that the Guru Academic Edition [30] and the model was setup with
traffic that traverses the routing paths exhibits. In [17-19] the help of many video and written tutorials. A network was
the traffic diversities are used incorporated using a relay created, consisting of 2 computers that exchange IP packets
routing mechanism for MP2P systems. Authors in [20-21] with one another through several routers that are in between
use a resource traffic offloading technique for energy usage them.
optimization in the cloud, using the centrality principle of A failure scenario also exists where the
social networks. Mobile users take advantage of the mobile communication between 2 routers would repeatedly fail and
opportunistic cloud, in order to increase their reliability in recover, in order to investigate the behavior of the routing
service provision by guaranteeing sufficient resources for protocols in such cases. Furthermore, the route each
the execution of mobile applications. In [22] authors apply protocol takes will be investigated by replacing the links of
mechanisms for precise performance analysis of reactive a particular route with faster ones, in order to check whether
routing protocols in Mobile Ad hoc Networks including the datagram packets will travel from the fastest or the
comparative evaluation of the efficiency under sudden shortest route to reach their destination.
failures. Very important comparative evaluations have been
exposed in [24-28] in both Cognitive radio level architecture
and heuristic routing decisions. In this work a comparison of A. Network model
four diversified routing decisions are presented exposing The created network is comprised of the following
both the pros and cons of each one the decisions/ components:
implementations.  Ethernet workstation: Workstation used with client-
server applications that runs over TCP/IP and
UDP/IP. It can support one underlying Ethernet
connection at 10Mbps, 100Mbps, or 1000Mbps and
B. Intermediate System to Intermediate System (IS-IS) for each packet, it requires a fixed amount of time in
IS-IS is a dynamic routing protocol designed and order to route them, as the "IP Forwarding Rate"
used for forwarding traffic within a packet-switched attribute of the node specifies. Packets are routed on
network by determining the best route for datagrams to take. first-come-first-serve basis and, depending on the
It is also based on the Dijkstra algorithm, and similarly to transmission rates of the corresponding output
OSPF, IS-IS is also an Interior Gateway Protocol (IGP), interfaces, they may encounter queuing at the lower
which uses link-state routing by reliably flooding link state protocol layers,.
information throughout a network of routers. Each  Ethernet64 switch: Switch that implements the
Spanning Tree Algorithm in order for the network
Intermediate System (IS or router), that is configured to run
topology not to contain any loops. It supports up to
IS-IS, creates a database of the network's topology
64 Ethernet interfaces. The communication between
aggregating the information[23]. switches is done by sending Bridge Protocol Data
 Units (BPDUs) to each other, and the packets are Traffic dropped: The number of IP datagrams
received and processed based on the current dropped by all nodes in the network across all IP interfaces.
configuration of the spanning tree.
 Cisco 7000 router: An IP-based router gateway
model. According to the destination IP address,
packets that arrive on an IP interface are routed to
the appropriate output interface.
 10BaseT duplex link: Ethernet connection link that
transmits data at a rate of 10Mbps and is able to
connect any Stations, Hubs, Bridges, Switches and
LAN nodes with one another.
 100BaseT duplex link: Similar to the 10BaseT
duplex link. The only difference is that the speed of
data transfer over the 100BaseT duplex link is
100Mbps instead of 10Mbps.

Figure 3: Average traffic dropped by the four protocols

As it is shown from Figure2, the protocol with the

most traffic drops in case of a link failure is OSPF and then
EIGRP. IS-IS doesn't drop many packages while RIP
doesn't drop any.

Network convergence activity: Records square

wave alternating between ordinates 0 and 1. While the there
are no changes in the forwarding tables in the network, the
Figure 2: Topology of the network value is 0. On the other hand, when some forwarding tables
show signs of convergence, meaning there has been a
The duration of the simulation is set to be 15 minutes with change in the tables, the value becomes 1.
scheduled failures between the routers R1 and R2. The
schedule of the failures is as follows:

Time Status
(seconds)
225 Fail
400 Recover
535 Fail
590 Recover
605 Fail
620 Recover
625 Fail
630 Recover
730 Fail
830 Recover Figure 4: Average Network Convergence Activity
Table1: Timing of failures/recoveries While EIGRP, IS-IS and RIP have almost the same
network convergence activity, for even less than a minute,
B. Results OSPF shows dramatic difference when compared to the
other 3 protocols, in both magnitude and longevity.
As mentioned above, the protocols that are
compared are the Routing Information Protocol (RIP),
Background Traffic: Measure of the end to end delay,
Enhanced Interior Gateway Routing Protocol (EIGRP),
experienced by information about background traffic flow
Open Short Path First (OSPF) and Intermediate System to
while it travels between the flow's source and the flow's
Intermediate System (IS-IS).
destination.
 will keep using the same route, by investigating the number
of hops from one PC to the other.

Figure 5: Average background traffic

Figure 7: New topology with faster links on the highlighted
In the case of background traffic, we can tell from part
Figure4 that IS-IS has the less traffic while the other 3
protocols have more or less the same. Number of hops: This statistic gives the average
number of IP hops taken by data packets reaching at a
To continue measuring, IP Unicast traffic was destination node.
created with intensity of 100Packets/sec, 120000Bits/sec
and 3600sec. of duration. The following measurements were
taken (the results of the previous simulation did not change
since it was simply an exchange of IP datagrams):

Delay: This statistic represents the end-to-end

delay of all packets received by all the stations.

Figure 8: Average number of hops before the link

replacement

Figure 6: Average packet end-to-end delay

Figure 5 shows that the most delay was caused by

RIP. While the other protocols had more or less the same
amount of delay, EIGRP seems to be the fastest one.

Lastly, as mentioned before, some of the 10BaseT

links were replaced with 100BasetT ones. As you can see
from the network topology, the shortest path from PC1 to Figure 9: Average number of hops after the link replacement
PC2 is no doubt PC1 -> Switch1 -> Router6 -> Switch2 ->
PC2. For this reason, the highlighted links were replaced We can see from the 2 figures that when all the
with faster links in order to see whether or not the protocols links were 10BaseT, the average number of hops the
different protocols would take was between 1.5 and 2.2
(Figure 1), which indeed indicates to the shortest path of Technology, 2010. Available at:
aforementioned. When the 10BaseT links were replaced http://www.bth.se/com/mscee.nsf/attachments/4983_Thesis_
Report_pdf/$file/4983_Thesis_Report.pdf
with 100BaseT ones (Figure 2), we can see that the average [5] Cisco Systems Inc., "OSPF Network Design Solutions", Cisco
number of hops for the protocols was increased to between Press Publications. Available at:
2 and 3.5 (OSPG and EIGRP overlap in both cases. That's http://docstore.mik.ua/cisco/pdf/OSPF%20Network%20Desig
why OSPF is not visible from the figures). n%20Solutions.pdf
With this information we can deduct that all the [6] J. Moy, "Multicast Extensions to OSPF", 1994. Available at:
http://tools.ietf.org/html/rfc1584.html
protocols chose to follow the faster route even if it meant
[7] G. Mastorakis, A. Bourdena, C. X. Mavromoustakis, E. Pallis, & G.
taking the more length one, rather than taking the shortest Kormentzas, (2013, July). An energy-efficient routing protocol for
and slower route. ad-hoc cognitive radio networks. In Future Network and Mobile
Summit (IEEE FutureNetworkSummit), 2013 (pp. 1-10).
[8] C. X. Mavromoustakis, A. Bourdena, G. Mastorakis, E. Pallis, & G.
V. CONCLUSION Kormentzas, (2014). An energy-aware scheme for efficient spectrum
utilization in a 5G mobile cognitive radio network architecture.
After the series of measurements, we can see that Telecommunication Systems, 1-13.
in case of failure, the best protocols are RIP and IS-IS. Of [9] C. X. Mavromoustakis, C.D. Dimitriou, & G. Mastorakis, On the
course RIP is not suitable for large networks since it Real-Time Evaluation of Two-Level BTD Scheme for Energy
supports only 15 hops. It is indeed the best in the case of Conservation in the Presence of Delay Sensitive Transmissions and
Intermittent Connectivity in Wireless Devices. International Journal
traffic drops but IS-IS is superior, even though its On Advances in Networks and Services, 6 (3 and 4), 148-162, 2013.
simulations were worse (but very close) to RIP, because of [10] C. X. Mavromoustakis, C. Dimitriou, G. Mastorakis, E. Pallis, "Real-
the fact that it supports larger networks. Time Performance Evaluation of F-BTD scheme for optimized QoS
When the average network convergence activity Energy Conservation in Wireless Devices", in Proceedings of the
was measured, all the protocols showed more or less the GlobeCom 2013 2013, pp. 1156-1161.
same behavior, except from OSPF, which showed much [11] O. Shiakallis, C. X. Mavromoustakis, G. Mastorakis, A.
Bourdena, E. Pallis, E. Markakis, C. Dobre, "A Scheduling
more activity than the other three protocols. Scheme for Throughput Optimization in Mobile Peer-to-Peer
For the average background traffic, IS-IS was the Networks", in the Book on "Emerging Innovations in
most efficient with the least background traffic, compared to Wireless Networks and Broadband Technologies",Editor:
Prof. Naveen Chilamkurti, publisher: IGI-GLOBAL, 2015.
the RIP, OSPF and EIGRP protocols that gave similar
[12] C. X. Mavromoustakis, P. Mousicou, K. Papanikolaou, G.
results to each other. Mastorakis, A. Bourdena, E. Pallis, "Dynamic cloud resource
In regards of the average packet end-to-end delay, migration for efficient 3D video processing in mobile
the protocol that did absolutely worse than the rest was the computing environments", Book on Novel 3D Media, by
RIP protocol, and while the results of the other three were Ahmet Kondoz and Tasos Dagiuklas, Novel 3D Media
Technologies, 2015, pp. 119-134, Springer International
close, the one that showed more efficiency was EIGRP. Publishing, January 2015.
Finally, when it came to making a routing decision [13] S. Mirtchev, C. X. Mavromoustakis, R. Goleva, K. Kassev, G.
between the shortest and fastest path, all the protocols chose Mastorakis, “Generic IP Network Traffic Management from
to follow the path that would transfer the datagrams faster Measurement through Analyses to Simulation” in the book
from the source to the destination, in spite of the number of “Resource Management in Mobile Computing
Environments”, pp. 167-192, 2014, Springer International
intermediate routers that they had to go through. Publishing.
[14] C. X. Mavromoustakis, A. Bourdena, G. Mastorakis, E. Pallis,
G. Kormentzas, and C. D. Dimitriou, "An Energy-Aware
Scheme for Efficient Spectrum Utilization in a 5G Mobile
REFERENCES Cognitive Radio Network Architecture", accepted for
publication in the Special Issue on Energy Efficient 5G
Wireless Technologies, Springer Telecommunication Systems
[1] T. H. Cormen, C. E. Leiserson, R. L. Rivest and C. Stein, Journal, 2015.
Introduction to Algorithms second edition, Chapter 24.1: The [15] A. Andreou, C. X. Mavromoustakis, G. Mastorakis, A.
Bellman-Ford algorithm and Chapter 24.3: The Dijkstra Bourdena, E. Pallis, “Adaptive Heuristic-Based P2P Network
algorithm, The MIT Press, McGraw-Hill Book Company, Connectivity and Configuration for Resource Availability” in
2001. the book Resource Management in Mobile Computing
Environments, pp. 221-240, 2014, Springer International
[2] D. Pei, D. Massey and L. Zhand. "A Formal Specification for Publishing.
RIP Protocol", Colorado State University, UCLA. Available [16] A. Bourdena, C. X. Mavromoustakis, G. Kormantzas, E.
at: ftp://www.mmss.cs.ucla.edu/tech-report/2004- Pallis, George Mastorakis, Muneer Bani Yassein, “A
reports/040046.pdf Resource Intensive Traffic-Aware Scheme using Energy-
[3] B. Albrightson, J. J. Garcia-Luna-Aceves and Joanne Boyle, efficient Routing in Cognitive Radio Networks”, Future
"EIGRP-A FAST ROUTING PROTOCOL BASED ON Generation Computer Systems Journal, Elsevier; Volume 39,
DISTANCE VECTORS", Cisco Systems, University of October 2014, pp. 16-28.
California. Available at: [17] C. X. Mavromoustakis, G. Mastorakis, A. Bourdena, E. Pallis,
https://www.ida.liu.se/~TDTS02/papers/eigrp.pdf “Energy Efficient Resource Sharing using a Traffic-oriented
[4] M. Nazrul Islam, M. A. U. Ashique, "Simulation Based Routing Scheme for Cognitive Radio Networks”, IET
EIGRP over OSPF Performance Analysis", Blekinge Institude
 Networks Journal, Volume 3, Number 1, March 2014 (IEEE Traffic Difference Estimation in Cognitive Radio Networks",
DL), pp. 54-63. in Proc. 14th IEEE International Symposium on a World of
[18] C. X. Mavromoustakis, G. Mastorakis, A. Bourdena, E. Pallis, Wireless, Mobile and Multimedia Networks (IEEE
“A Resource Intensive Traffic-Aware Scheme using Energy- WoWMoM 2013), 3rd IEEE Workshop on Convergence
Efficient Routing in Cognitive Radio Networks”, Journal of among Heterogeneous Wireless Systems in Future Internet
Information Technology Research, IGI-Global, 2015. (CONWIRE 2013), Madrid, Spain, 04-07 June, 2013, pp. 01-
06.
[19] C. X. Mavromoustakis, G. Mastorakis, A. Bourdena, E. Pallis,
Georgios Kormentzas and C. Dimitriou, "Joint Energy and [25] A. Bourdena, E. Pallis, G. Kormentzas, G. Mastorakis, “Radio
Delay-aware Scheme for 5G Mobile Cognitive Radio Resource Management Algorithms for Efficient QoS
Networks", to appear in proceedings GlobeCom 2014, track Provisioning over Cognitive Radio Networks”, in Proc. IEEE
Globecom 2014 - Symposium on Selected Areas in International Conference on Communications, IEEE
Communications: GC14 SAC Green Communication Systems ICC2013, Budapest, Hungary, 09-13 June, 2013.
and Networks - GC14 SAC Green Communication Systems [26] C. D. Dimitriou, C. X. Mavromoustakis, G. Mastorakis, E.
and Networks, Austin, TX, 8-12 December 2014, Austin, Pallis, "On the performance response of delay-bounded
Texas, USA, pp. 2665-2671. energy-aware bandwidth allocation scheme in wireless
[20] K. Papanikolaou, C. X. Mavromoustakis, G. Mastorakis, A. networks", in Proc. IEEE ICC2013, Workshop, Budapest,
Bourdena, C. Dobre, "Energy Consumption Optimization Hungary, 09-13 June, 2013.
using Social Interaction in the Mobile Cloud", in Proc. of [27] G. Mastorakis, C. X. Mavromoustakis, A. Bourdena, G.
International Workshop on Enhanced Living EnvironMENTs Kormentzas, E. Pallis, “Maximizing Energy Conservation in a
(ELEMENT 2014), 6th International Conference on Mobile Centralized Cognitive Radio Network Architecture”, in Proc.
Networks and Management (MONAMI 2014), Wuerzburg, 18th IEEE International Workshop on Computer-Aided
Germany, September 2014. Modeling Analysis and Design of Communication Links and
[21] G. Mastorakis, E. K. Markakis, E. Pallis, C. X. Networks (IEEE CAMAD 2013), Berlin, Germany, 25-27
Mavromoustakis, G. Skourletopoulos, "Virtual Network September, 2013.
Functions Exploitation through a Prototype Resource [28] G. Mastorakis, C. X. Mavromoustakis, A. Bourdena, E. Pallis.
Management Framework", International Conference on G. Sismanidis, “Optimizing Radio Resource Management in
Telecommunications and Multimedia TEMU 2014, IEEE Energy-Efficient Cognitive Radio Networks”, in Proc. 2nd
Communications Society proceedings, 28-30 July, Crete, ACM Workshop on High Performance Mobile Opportunistic
Greece, pp. 24-28. Systems (HP-MOSys 2013), 16th ACM International
[22] M. Papadopoulos, C. X. Mavromoustakis, G. Conference on Modeling, Analysis and Simulation of
Skourletopoulos, G. Mastorakis, E. Pallis, "Performance Wireless and Mobile Systems, in cooperation with R8 IEEE
Analysis of Reactive Routing Protocols in Mobile Ad hoc Computer Society C16, Barcelona, Spain, 03-08 November
Networks", International Conference on Telecommunications 2013.
and Multimedia TEMU 2014, IEEE Communications Society [29] CISCO official website:
proceedings, 28-30 July, Crete, Greece, pp. 104-110. http://www.cisco.com/en/US/products/ps6599/products_white
[23] H. Smit and T. Li, "Intermediate System to Intermediate _paper09186a00800a3e6f.shtml
System (IS-IS) Extensions for Traffic Engineering (TE), [30] OPNET official website:
2004. Available at: https://tools.ietf.org/html/rfc3784. http://www.opnet.com/university_program/itguru_academic_
[24] G. Mastorakis, C. X. Mavromoustakis, A. Bourdena, E. Pallis, edition/
"An Energy-Efficient Routing Scheme using Backward