Token based Supervisory System for mobile Ad hoc

Networks
1 2 3
Konyeha C. Konyeha S. Mughele E. S.
Department of Electrical/ Department of Computer Department of Computer
Electronic Engineering Science Science
Benson Idahosa University University of Benin, Benin University of Benin, Benin
Benin City, Nigeria City, Nigeria City, Nigeria
[email protected] [email protected]

Abstract
Mobile Ad hoc network (MANET) performance, is dependent on the participation of all network
nodes to forward packets efficiently due to its decentralized authority architecture. A node may refuse
to cooperate in order to save its resources while still using the network to relay its own traffic. This
selfish behaviour, if exhibited by many nodes, degrades network performance and cooperating nodes
become unfairly loaded. Existing detection and quarantining mechanisms have ensured that selfish
nodes do not receive network services, thus penalizing selfish nodes. However, if many nodes become
selfish, network communication itself becomes impossible. In this work, we introduce a hybrid incentive
with tolerance mechanism to the token-based umpiring technique (TBUT), an existing selfish node
detection and quarantining technique, in order to discourage selfishness amongst nodes and reduce the
impact of quarantining selfish nodes on the network performance. The simulation results show that our
method in principle enhances TBUT’ framework performance.
Keywords: MANET, packet forwarding, routing, collaborate, selfish node, misbehave, token based,
umpire, network performance

1.0 Background of Study

A mobile ad hoc network (MANET) is an infrastructure-less, dynamic topology network, consisting
of a collection of mobile nodes that communicate with each other without the use of any centralized
authority. These wireless nodes move about arbitrarily and can act as source, destination, or intermediate
routers in order to do both data transmission and reception in a network. The mobility of nodes causes
the network topology to be unpredictable and to change frequently. The absence of a centralized
authority introduces the need for nodes to cooperate and self-organize to form a working communication
network.
Each mobile node in a MANET consists of a transceiver which uses carrier sense multiple access with
collision avoidance (CSMA/CA) to achieve efficient communication among nodes. Communication
among nodes may be established directly or indirectly, a source node will have a direct communication
with the destination node that is within radio range. Nodes which are not within radio range will need
the cooperation of other nodes in the network, to forward their packets using multi-hop communication.
This results in an indirect communication amongst nodes (Aarti, 2013).
The flexibility of MANETs make it attractive for emergency situations such as military deployment
in a hostile environment and search-and-rescue operations. It is also used in areas where a
communication network is required, but there is none available, or the preexisting infrastructure has
been destroyed by a disaster or a war.
2.0 Review of related literature
Data transfer is a very important function in MANETs and requires that for any pair of non-adjacent
nodes, the intermediate nodes forward data packets. It is therefore assumed by routing algorithms that
every node will forward each packet it receives in a timely manner, but this is not always the case as
some nodes may make use of other node’s resources for communicating and refuse to forward packets
from other nodes within their radio range. These nodes are termed as selfish or misbehaving (Ciobanu
et al 2013). The most common cause of selfishness in nodes is the shortage of resources including battery
life, memory and processor power (Kampitaki et al, 2014). This results in the node saving its resources,
while disregarding the requests to forward packets to other nodes in the network. Furthermore, a node
could become selfish to stop malicious data packets from untrusted nodes (Ciobanu et al 2013). The
misbehavior of nodes as a result of their selfish reasons can significantly diminish the performance of
MANET (Kashif and Kadam, 2016).
The performance of a MANET is dependent on the collaboration of all nodes in the network to
forward packets for each other. The selfish behavior of nodes, if exhibited by many nodes, degrades
network performance and cooperating nodes become unfairly loaded. The detection and quarantining
mechanisms proposed in literatures have succeeded in ensuring that on detecting selfish nodes, they are
isolated from using any network services, thus penalizing the selfish node. However, simply detecting
and quarantining selfish nodes may not be enough to improve the performance of a network (Ciobanu
et al 2013).
Some of the characteristics of selfish nodes according to Gupta (2011) are as follows; selfish nodes
may not take part in packet routing process, selfish nodes may not be detected by the other nodes when
required, selfish nodes do not reply or send hello messages, selfish nodes exempt themselves from
participating in routing, or they may drop the data packets while relaying it to the next hop such that a
degradation in the performance of the routing path is observed by the source and destination, but they
are unable to detect the problematic link (Zhang, 2011).
Several techniques for selfish node detection have been proposed to address the selfish behavior in
nodes from the network perspective, these include the techniques that address packet dropping or refusal
to forward packets. The techniques are classified into three categories as stated in the work of Zhang,
2011. These are:
1. Reputation-based systems: Each node monitors the other nodes, observes their behavior and uses
the acquired information for routing.
2. Credit-based systems: Each node gives a credit to other nodes as a reward for forwarding data.
3. Acknowledgement-based systems: Each destination node sends an acknowledgement to prove that
it received the data packet forwarded by the source.
A summary of some of the related works reviewed is presented in Table 1.
Table 1: Summary of some related works
AUTHOR METHOD STRENGTH WEAKNESS

Enrique et al Proposed a collaborative This approach reduces the This technique is used for just
(2015) watchdog approach (CoCoWa), time and increases the detecting selfish nodes and there is no
which is based on the fast precision of selfish node penalty on such nodes after they are
diffusion of selfish nodes detection. detected.
awareness.
 Janakiraman Proposed an Erlang distribution Estimates the impact of Selfish nodes do not participate
and Rajendiran based Conditional Reliability selfish nodes on the further in any network activity and
(2016) Mechanism (ECRCM) that aids in resilience of the network this increases the workload on the
detecting and isolating the selfish and performs well as cooperating nodes.
nodes present in an ad hoc compared with other
environment selfishness
mitigating techniques
Soltanali et al. Proposed a scheme to encourage It does not require a Selfish nodes do not participate further
(2007) packet forwarding and discipline tamper resistant hardware in any network activity and this
selfish behavior in a non- in contrast to the tamper increases the workload on the
cooperative ad hoc network. A resistant hardware, which cooperating nodes. This scheme
punishment scheme is designed to is not applicable for a pure requires beaconing which may incur
punish selfish nodes. ad hoc network quite a bit of overhead
Kumar et al. Proposed a token-based umpiring It is very efficient with Selfish nodes do not participate
(2015) technique (TBUT) in which every reduced detection time and further in any network activity and
node needs a token to participate less overhead. It is this increases the workload on the
in the network activities and the predominant in enhancing cooperating nodes.
neighboring nodes acts as an the network performance
umpire. Umpire nodes will and improving the security
monitor the behavior of the nodes of most of the real
and detect if any node is application
misbehaving.

This work is based on the technique proposed by Kumar et al. (2015), the system employed the use
of tokens in its analysis to accelerate the detection and elimination of selfish nodes. Each node is issued
with a token at its inception. The token consists of three fields: NodeID, status, and reputation. NodeID
is assumed to be unique and deemed to be beyond manipulation; status is a single-bit flag. The status bit
and the reputation value are initialized to zero. The token with a green flag and positive reputation is a
permit issued to each node, which confers it the freedom to participate in all network activities. Each
node in order to participate in any network activity, has to announce its token status bit and reputation
value. If token status bit and reputation value are “1” and “-1” respectively, the protocol does not allow
the node to participate in any network activity. The simulation results for TBUT has shown that the
technique is efficient for packet routing but can give improved performance by minimizing the penalty
to innocent node.
As a result of the aforementioned, this research therefore faults TBUT on its unfair quarantining of
selfish nodes immediately they are detected. Also, if all the nodes in a MANET are quarantined, then
we might at some point not have a network communication as all nodes have the tendency to be selfish.
Therefore, this work aims to improve the performance of TBUT by employing an incentive
mechanism, in order to encourage cooperation and discourage selfishness among nodes in a MANET,
and modifying TBUT by introducing a tolerance technique in order to reduce the impact of selfish node
quarantining on the network, thus leading to a more robust and reliable network performance.
3.0 Methodology
The benefit of an efficient MANET can only be obtained by nodes if they cooperate, if a node refuses
to forward other nodes’ packets, this results in the node packets being forwarded to destination node
while refusing to forward packets from other nodes to their destination nodes, if all nodes refuse to
forward data packets then no network is formed.
Figure 1: A representation of TBUT model (Kumar et al, 2015)

Figure 2: Developed system model

3.1 Algorithm for Selfish Node Detection and Quarantining
Begin
1. //Route Discovery
2. Step 1: Source (S) sends a RREQ to Destination (D)
3. Step 2: Destination (D) receives the RREQ from source (S) and send RREP to the source (S)
4. Step 3: Umpiring node are appointed to the corresponding neighboring nodes in the network.
5. //Reputation Calculation
6. Step 4: Neighbor information is gathered and sensed.
7. Step 5: Cross checking with the neighbour reputation information and their own reputation calculation.
8. If (own reputation calculation value == neighbour reputation information)
9. {
10. If (selfish node is detected)
11. Set S_Count = S_Count + 1
12. If (S_Count ==2)
13. Set flag bit “1” and reputation value bit “-1”
 14. Add selfish node to the selfish block list (SBL)
15. Else
16. Reputation value is incremented by 1
17. Transfer the data to its corresponding destination node;
18. }
19. Step 6: Finally, the performance is measured

3.2 Incentive Mechanism

A selfish node will need incentive in order to forward other node’s packets since doing so will incur
cost (of energy and other resources). The incentive mechanism ensures that messages from defaulting
nodes are not accepted, thus forcing them to participate if they want their messages delivered. Therefore,
whenever a node B has its S_Count as 1, it is set on probation and is considered selfish by node A and
the other two umpires, it is notified, and node C won’t accept any messages from it (but node A will
keep on forwarding its own messages to node B). Therefore, a selfish node might end up not being able
to send its messages, unless it becomes unselfish.
4.0 Performance testing using Network Simulator (NS-2)
This research made use of NS-2 for simulation due to its large number of available realistic mobility
models, powerful scripting, simulation setup, and most importantly, its large number of user community
for technical support. Qualnet 5.0, an event driven network simulator was used to implement the TBUT
framework, but due to its unavailability in the commercial-off-the-shelf market despite the fact that it is
a commercial software product as stated by its vendors, we were left with the choice of simulation using
the next best tool that has the capacity to support the simulation of our ideas. All efforts to get a hold of
the Qualnet simulator proved futile as our e-mails to the vendors, and the authors of the existing system
were never replied.
NS-2 is a network simulator which provides support for simulation of Transmission Control Protocol
(TCP), routing, and multicast protocols over wired and wireless (local and satellite) networks
(Bhuvaneshwari, 2016). It is based on C++ which is used to define the behavior of protocols, and OTcl
(Object Oriented Tool Command Language), which is used for scenario configuration, manipulation of
existing C++ objects, periodic or triggered actions. NS-2 uses OTcl programming language to interpret
user simulation scripts. The Tcl language is fully compatible with the C++ programming language. At
the top layer, NS-2 is an interpreter of Tcl scripts of the users.

Figure 3: NS-2 Schema (Sheikh,2012)

NS-2 has the NAM (Network Animator) object which is a Tcl based animation tool for viewing
network simulation traces, and it shows the visual animation of the simulation. Another component is
the trace object that contains history of the behavior of all objects in the simulation. They are both
created as files by NS-2. The former is a .nam file used by NAM software that comes along with NS-2.
Figure 4 shows the Nam environment consisting of 100 mobile nodes in NS-2. The latter is a “.tr” file
that includes all the simulation traces in the text format. NS project is normally distributed along with
various packages (ns, nam, tcl, otcl etc.) named as “all-in-one package”, but they can also be found and
downloaded separately. In this study we have used version 2.35 of NS all-in-one package and installed
the package in the Linux environment. The Tcl script was written in a text editor (gedit) and the linux
command, “ns filename.tcl” was used to initiate the simulation.

Figure 4: NAM environment of NS-2

4.1 Performance Test Simulation Parameters
Our simulation parameters are set as shown in Table 2.
Table 2: Parameter setting
Simulation Parameters VALUES
Simulation time 1000s
Transmission range 250m
Bandwidth 2Mbps
Mobility model Random way point
Propagation model Two-ray ground reflection
Maximum speed 0 – 20 m/s
Pause time 0s
Traffic type Constant Bit Rate (CBR)
Payload size 512 bytes
Simulation Area 3000m x 1600m
Routing protocol AODV
Number of mobile nodes 100

4.2.1 Implementation of the Selfish Node Behaviour

Though NS 2.35 contains the routing protocol (AODV) we require for this research, it does not have
any modules to simulate selfish behavior in nodes. To this effect, we modified the AODV routing
protocol file using C++ in the NS 2.35. We implemented the selfish node behaviour by editing the
aodv.cc and aodv.h file whereby the selfish node will simply drop the packet, then recompiled NS-2
files to create object. Having finished compilation, we have a new test bed to simulate selfish nodes in
AODV protocol.
4.2.2 Implementation of Selfish Node Detection and Quarantine
In this work, we achieved selfish node detection by setting the nodes in promiscuous mode, where each
node in the network listens for packets transmitted by its neighbour nodes and takes action in the event
of a packet drop. To accomplish this, we modified three files; aodv.cc, aodv.h and ns-mobilenode.tcl.
In our simulation, we observed that introducing the S_count field reduced the number of quarantined
nodes in the 1000s simulation time. This helped to reduce the impact of selfish node quarantining which
is implemented by placing a selfish node in a block list. However, it is not possible to implement a
block list in NS2, however we were able to achieve our idea by moving the selfish node out of radio
range so they are not able to communicate.
4.2.3 Implementation of the Tolerance and Incentive Mechanism
The hybrid tolerance with incentive mechanism was implemented by modifying the token. We added
to the token of each node, an S_count field which serves the dual purpose of reducing node quarantine
and discouraging selfishness among nodes by punishing non-cooperating nodes.
4.3 Performance Metrics
The presence of selfish nodes in the network decreases the throughput and increases the total and control
overhead (Sandhini et al, 2016). Hence the developed technique is analyzed based on its packet delivery
ratio.
4.3.1 Packet Delivery Ratio
This is the ratio of the number of packets received by destination nodes to the number of packets actually
sent to it by source nodes. It is accepted as a standard measure of throughput. Figure 6 shows the packet
delivery ratio in the presence of 30% selfish nodes. Five different files were created to obtain the packet
delivery ratio for mobility from 0 to 20. We generated the throughput for the different mobility trace
files as shown in figure 5, using an awk file with the command;
awk –f genthroughput.awk filename.tr
Figure 5: Computing the throughput for 0 – 20m/S Mobility

Figure 6: Packet delivery rates, for 30% selfish node with node mobility varying between 0 - 20 m/s.

4.4 Throughput of Generating Packets

The graph plotted in Figure 4.5 shows the throughput of forwarded packets on y-axis versus the
simulation time on x-axis. We see an increase in throughput at simulation time 60.

Figure 7: Throughput of generating packets vs simulation time

4.4.1 Throughput of Dropping Packets
Figure 8 shows the graph of the throughput of dropping packets on the y-axis and the simulation time on the x-
axis.
Figure 8: Throughput of dropping packets vs simulation time

4.5 Result and Discussion

The packet delivery ratio decreases as mobility of selfish nodes increases. In the case of TBUT, with
30% selfish nodes, the packet delivery ratio decreases from 76.83%, when the nodes are stationary to
64.98%, when the nodes are moving at 20m/s. In ETBUT, the packet delivery ratio varied for different
mobility scenarios and went from 70.90% at 0 mobility to a 57.64% at 20m/s. This indicates that as
compared to ETBUT, the packet delivery ratio of TBUT are superior. We still believe in the efficiency
of ETBUT as considering the problems it mitigates, it is a more efficient and secure security scheme for
MANET.
5.0 Conclusion
Presently, most literatures available on selfish node detection and prevention focus more on detecting
and eliminating selfish nodes immediately they drop packets. This research faults this elimination
approach as so many reasons exist for this packet dropping in nodes some of which may be legitimate.
Also, every node in MANET has the tendency of being selfish and quarantining or eliminating them on
the spot without giving them a second chance might at some point result in degradation in network
performance or altogether, no network communication.

References
Aarti, D. S. (2013). Study of MANET: Characteristics, Challenges, Application and Security Attacks.
International Journal of Advanced Research in Computer Science and Software Engineering.
Bhuvaneshwari, R. K. (2016). Improving Selfish Node Detection n MANET Using A Collaborative
Watchdog. International Journal of Advanced Research Trends in Engineering and Technology
(IJARTET), Vol. 3, No. 2, pages 125-134.
Ciobanu, R., Dobre, C., Dascăl, M., Trăuşan-Matu, S., and Cristea, V. (2014). SENSE: A Collaborative
Selfish Node Detection and Journal of Network and Computer Applications. Journal of Network and
Computer Applications, Volume 41, May 2014, Pages 240-249
Enrique H., Olmos, M. D. S., Cano, J., Calafate, C. T., and Manzoni, P. (2015). CoCoWa: A
Collaborative Contact-Based Watchdog for Detecting Selfish Nodes. IEEE Transactions on Mobile
Computing (Volume: 14, Issue: 6 , June, 2015, pp. 37 - 42
Gupta, S. (2011). Impact of Selfish Node Concentration in MANETs. International Journal of Wireless
& Mobile Networks (IJWMN).
Janakiraman, S., and Rajendiran M. (2016). An Erlang Factor-Based Conditional Reliability Mechanism
for Enforcing Co-operation in MANETs. Serbian Journal of Electrical Engineering, Vol. 13, No. 2,
June 2016, 265-284
Kashif, M., and Kadam, P. V. (2016). Improved collaborative watchdog system for detection of selfish
node in MANET. International Journal of Science, Engineering and Technology Research (IJSETR).
Kumar, A. K., Jebakumar, M., and Singh, P. J. (2015). A unified approach for detecting and eliminating
selfish nodes in MANETs using TBUT. EURASIP Journal on Wireless Communications and
Networking, Volume 10, No. 30, June 2016, pages 32–37.
Sandhini, S. S. (2016). False Alarm Method for Detecting Selfish Node in Manet. International Journal
of Future Generation Communication and Networking.
Soltanali, S. (2007). In Proceedings of the Third International Conference on Networking and Services.
In Proceedings of the 10th IEEE Symposium on Computers and Communications (ISCC07), (pp. 37
- 42).
Zhang, Y. (2011). Detection and Isolation of Packet Droppers in Wireless Ad-Hoc Networks. IEEE
Transactions on Wireless and Cellular Networks, VOL. 51, NO. 4, April 2011, 1 - 37.