GEOGRAPHIC INFORMATION-BASED ROUTING OPTIMIZATION USING GA FOR CLUSTER-BASED WSNS

International Journal of Wireless & Mobile Networks (IJWMN) Vol. 10, No. 3, June 2018
DOI: 10.5121/ijwmn.2018.10302 13
GEOGRAPHIC INFORMATION-BASED ROUTING
OPTIMIZATION USING GA FOR CLUSTER-BASED
WSNS
Sanghyeok Lim1
and Taeho Cho2
1
College of Information and Communication Engineering, Sungkyunkwan University,
Republic of Korea
2
College of Software, Sungkyunkwan University, Republic of Korea
ABSTRACT
Wireless sensor networks are used for data collection and event detection in various fields such as home
networks, military systems, and forest fire monitoring, and are composed of many sensor nodes and a base
station. Sensor nodes have limited computing power, limited energy, are randomly distributed in an open
environment that operates independently, and have difficulties in individual management. Taking
advantage of those weaknesses, attackers can compromise sensor nodes for various kinds of network
attacks. Several security protocols have been proposed to prevent these attacks. Most of the security
protocols form routings with cluster head nodes. In the case of routing using only cluster head nodes, it is
difficult to re-route when the size of the cluster is increased or the number of the surviving nodes is reduced.
To prevent these attacks, the proposed scheme maintains security in a cluster-based security protocol and
shows energy efficient routing using genetic algorithm by selecting the appropriate cluster head nodes and
utilizing the characteristics of the sensor node with different transmission outputs based on the distance
between each node. In this paper, we use a probabilistic voting-based filtering scheme, one of the cluster-
based security protocols, and the shortest path, which is a hierarchical routing protocol that the original
probabilistic voting-based filtering scheme is using, to test the proposed scheme. This experiment shows the
performance comparison of the routing success rate and routing cost according to the number of nodes on
the field, as well as the performance comparison according to the cluster size per number of nodes.
KEYWORDS
Network Protocols, Wireless Sensor Network, Network attack, Genetic algorithm, Routing
1. INTRODUCTION
Wireless sensor networks (WSNs) are used for data collection and event detection in various
fields, such as home networks, military systems, and forest fire monitoring, and are composed of
many sensor nodes and a base station (BS) [1], [2]. When an event occurs, the sensor node detects
the event, makes a report of that event and sends it over multiple hops of the sensor nodes to the
BS. However, sensor nodes are vulnerable to attack because of the disadvantages of limited
computation, limited energy, random distribution in an open environment that operates
independently, and difficulties in individual management. An attacker exploits these weaknesses
to compromise nodes and perform various kinds of attacks. These attacks reduce the energy of the
node, shorten the lifetime of the network, and prevent the detection and reporting of normal
events. Several protocols have been proposed to increase the security and lifetime of the network
by defending against such attacks [3], [4], [5], [6], [21]. These protocols form cluster-based
routing, which benefits from local management of nodes and it makes systemic report generation
possible. However, such cluster-based routing, which relies only on CH nodes for report
transmission and verification, has a significant negative impact on security considering the entire

International Journal of Wireless & Mobile Networks (IJWMN) Vol. 10, No.
network because of various attacks
downstream event detection cluster
most WSN fields, the node with the shortest distance to the BS and within the transmission range
based on the report delivery node is selected as the next transfer node.
of nodes is appropriately placed in the field
reasonable cost.
However, if the number of nodes placed in the field is insufficient or the size of the cluster is
inevitably increased due to a change
is likely to be reduced, and it becomes difficult to guarant
increases the total number of routing hops, thereby reducing the lifetime of the network.
to prevent such problems, the CHs of each event detection cluster perform
report generation, and report transmission.
among all the nodes with an appropriate geographical advantage considering the distance to the
BS and transmission power. The existing layer
security protocols implement routing
distance between nodes [7], [8]
different transmission output levels according to
the routing is constructed using
setting the appropriate output level for all nodes forming the routing.
distance of the receiving node from the initial event detection node, the
the node with the smallest distance from the BS
existing protocol randomly select the CH or select the n
experiments, we confirmed that the performance difference between the proposed scheme and the
existing scheme increases as the number of nodes decreases.
scheme based on PVFS which is
routing.
2. RELATED WORK
2.1. PVFS
International Journal of Wireless & Mobile Networks (IJWMN) Vol. 10, No. 3, June
attacks, such as sniffing or dropping reports transmitted from the
event detection cluster. Additionally, considering the shortest path that is applied to
, the node with the shortest distance to the BS and within the transmission range
based on the report delivery node is selected as the next transfer node. When a sufficient number
ropriately placed in the field, this ensures stable routing formation and
change in the field environment, the forwarding node candidate list
it becomes difficult to guarantee energy efficient routing
increases the total number of routing hops, thereby reducing the lifetime of the network.
to prevent such problems, the CHs of each event detection cluster perform only event detec
and report transmission. The verification node is considered
an appropriate geographical advantage considering the distance to the
The existing layer-based routing schemes applied to most
security protocols implement routing sets maximum transmission output without regard to the
], [8]. Micaz mote used in most of the security protocols
different transmission output levels according to each node’s distance [9]. In the proposed scheme,
the routing is constructed using genetic algorithm (GA) to minimize the cost of the routing by
setting the appropriate output level for all nodes forming the routing. In addition, to minimize the
ing node from the initial event detection node, the CH node was selected as
the node with the smallest distance from the BS, thereby saving the initial routing cost while
randomly select the CH or select the node with the smallest ID valu
existing scheme increases as the number of nodes decreases. We experimented
is one of a cluster-based security protocol using the
Figure 1. PVFS’s en-route filtering.
e 2018
14
or dropping reports transmitted from the
the shortest path that is applied to
, the node with the shortest distance to the BS and within the transmission range
a sufficient number
ensures stable routing formation and a
forwarding node candidate list
routing. This routing
increases the total number of routing hops, thereby reducing the lifetime of the network. In order
event detection,
The verification node is considered and selected
an appropriate geographical advantage considering the distance to the
applied to most of the
without regard to the
protocols can set
In the proposed scheme,
to minimize the cost of the routing by
In addition, to minimize the
node was selected as
the initial routing cost while
ode with the smallest ID value. Through
our propose
shortest path

15
PVFS is a typical cluster-based security protocol. Nodes randomly placed on the field are
assigned to the same cluster within an appropriate range to detect events within the same cluster
area. At the end of the node deployment, PVFS performs network security and report
transmission in four steps.
Key distribution phase
The BS transmits to each CH a set of keys separated by a cluster size. The CH that receives the
key set distributes one key to each member node.
Selection step of the verification node
The verification node is selected by the probability based on the number of hops between the
event detection cluster and the BS, and the number of hops between the node and BS among the
CHs on the path. The CH selected as the verification node receives one of the keys of the member
nodes of the event detection cluster at random. The CH of the selected cluster in the verification
node clears all keys except its own in its own key set.
Report delivery phase
After detecting the event, the CH generates a report on the event, receives the MAC created by
the member nodes of each node, and attaches it to the report to generate the final report.
Verification step
When the verification node receives the report, it first compares its own key index with the key
index of the MAC attached to the report. If the key index overlaps, the MAC check is performed.
If the index does not overlap, the report is not verified and sent to the node on the next path. If a
normal MAC is detected as a result of the check, a normal vote is cast. If a false MAC is detected,
a false vote is cast. If the number of votes reaches a preset threshold value, the report is
discriminated as normal or false. A report that is deemed to be a normal report is then
immediately directed to the BS without any further verification steps.
2.2. Genetic Algorithm
The genetic algorithm (GA) was proposed by Holland John and takes ideas from evolutionary
processes occurring in the natural world. This kind of algorithm represents solutions to a problem
as bit strings and finds an optimal solution through evolution [10], [22], [23]. The GA unit
probabilistically selects two individuals constituting the current generation. The selection
probability for each individual is proportional to the fitness of the individual, and the child
generations consist of individuals with greater fitness than the members of the parent generation.
Selected individuals generate a new generation through child generating, mutating, and
sometimes elite choosing [11], [12]. To solve the problem of local fixation in evolution the
proposed GA uses mutation technic.
3. PROPOSED METHOD
3.1. Problem statement
3.1.1 Routing problems
A WSN with a vast land area has few obstacles between deployed nodes, and cluster-based
security protocols mainly use the shortest path based routing. In a highly selective environment,

16
this is efficient for the forwarding node in the routing formation. However, in many security
protocols, including PVFS, the selection range of the forwarding node is limited to the CH among
all nodes. Moreover, once the CH is selected, it will not be replaced by any other member node.
In the situation where the CH is being used for extortion or because of battery discharge, the
corresponding routing scheme shows a significant weakness in re-routing. In some cases, when
the next forwarding node is set based only on the distance to the BS, the selection of the
geographically optimized node cannot be performed, which results in the adverse effect of the
increased number of routing hops.
3.1.2 Delivery node selection problem within cluster-based security protocols
Most of the cluster-based protocols select only those nodes that are responsible for report delivery
among the CH nodes. There are two bottlenecks in the routing path: key storage and energy
consumption. The key storage bottleneck is a phenomenon in which the number of keys owned
by the CH for the verification of downstream nodes increases. When these CHs are compromised,
the attacker will have multiple keys and can cause a significant security problem. The bottleneck
of energy consumption is the phenomenon that the routing of the downstream nodes to the BS
proceeds through the corresponding CH. As the number of reports to be transferred to the CH
increases exponentially, the expectation of energy consumption can be greater than other nodes
and the battery will be discharged soon. The network will then have to go through the routing
again and additional energy consumption is expected.
3.1.3 Uniform output power of sensor nodes
In the existing security protocols, all nodes transmit the report to its upstream nodes with a
maximum transmission range. In actuality, it is unnecessary to communicate only with the
maximum transmission distance by disregarding the interval between each sensor node. It is
therefore useful to measure the distance between nodes and transmit them at the appropriate
output level. The distance measurement between nodes can be implemented using the RSSI value
[13], or the node attached GPS module [14]. The Micaz node, which is often used in WSNs, is a
node that can control the output of a total of 32 levels, allowing the user to set the appropriate
output value [15], [16], [17]. There are many experiments on the transmission distance according
to the output in actual situations [18], [19], [20].
3.2. System overview
The experiment of the proposed method constitutes the system based on PVFS. All key
distribution and en-route filtering procedures follow existing security protocols. After the node is
deployed, each node sends its location information to the BS. Based on this information, the BS
selects the nodes with the shortest distances from the BS itself in each cluster as CH. The
difference in geographical location in the selection of CH has no effect on the security of the
protocol. This is because in the existing security protocol, CH selection was made without taking
any other factors into account. The key distribution and verification node selection phase follows
the original protocol’s phase. Then, the intermediate delivery node is randomly selected among
all the nodes, including the cluster head node. The selection of the next forwarding node is based
on the nodes within the maximum transmission range of itself and re-divides the range based on
the distance to the BS among the nodes within its transmission range.

Figure 2. Range of the candidate list of the forwarding node
In Figure 2, (a), (b), and (c) are the range
nodes can be selected by the downstream
diversity of the nodes forming the routing.
information overlapping between parent chromosomes in
a next generation node having a cross
range of the forwarding node will be explained
ange of the candidate list of the forwarding node.
and (c) are the ranges of the candidate list of the forwarding node.
downstream node with a larger range, resulting in an increase in the
the nodes forming the routing. However, this means that there is less genetic
information overlapping between parent chromosomes in the GA, and it is also difficult to create
a cross-point between parents. The contents of the candidate list
will be explained after dealing with the routing through the GA.
e 2018
17
of the candidate list of the forwarding node. More
, resulting in an increase in the
However, this means that there is less genetic
difficult to create
The contents of the candidate list
after dealing with the routing through the GA.

Figure
The GA chromosome contains the IDs of the forwarding nodes
chromosome is equal to the number of hops in each routing
is inversely proportional to the final
of the two parental chromosomes selected do not overlap, they will not have a cross
case, the parent chromosome is re
parent chromosome is large, the probability that the genetic information of each other does not
overlap increases. The proposed scheme limits the re
genetic information overlaps. When the number of re
chromosomes with the highest fitness are re
a modified version of the elitism technique used in genetic algorithms.
changed the size of the node list to increas
Additionally, we did not use the elitism technique
of the candidate list of the transfer node of
chromosome generating ability of the GA and
to re-select the routing within a short time, the scope of the candidate list can be narrowed and the
number of GA routines can be reduced to form an immediate routing
greater. Conversely, if the user wants to optimize the routing cost,
of the list and increase the number of repetitions of the GA.
Figure 3. Routing optimization using the GA.
The GA chromosome contains the IDs of the forwarding nodes, and the length of each
number of hops in each routing path. The fitness of the chromosomes
is inversely proportional to the final routing cost of the chromosomes. If the genetic information
of the two parental chromosomes selected do not overlap, they will not have a cross-
case, the parent chromosome is re-selected through the same method. If the candidate to be the
The proposed scheme limits the re - election of parental chromosomes until
When the number of re-elections reaches a predetermined threshold,
chromosomes with the highest fitness are re-elected and transferred to the next generation
a modified version of the elitism technique used in genetic algorithms. Instead of
list to increase the diversity of chromosomes.
, we did not use the elitism technique that causes the fixation phenomenon.
of the candidate list of the transfer node of Figure 2 has a trade-off between the child
ability of the GA and its routing cost. Therefore, if the network user has
select the routing within a short time, the scope of the candidate list can be narrowed and the
number of GA routines can be reduced to form an immediate routing, even if the cost is slightly
Conversely, if the user wants to optimize the routing cost, the user can increase the range
number of repetitions of the GA.
e 2018
18
and the length of each
The fitness of the chromosomes
If the genetic information
-point. In this
If the candidate to be the
election of parental chromosomes until
hes a predetermined threshold,
elected and transferred to the next generation. This is
Instead of mutation, we
causes the fixation phenomenon. The size
off between the child
Therefore, if the network user has
select the routing within a short time, the scope of the candidate list can be narrowed and the
if the cost is slightly
can increase the range

Figure 4 shows the original routing
Route # 2 have two cross-points. These are routing information for
respectively. Thus, the child chromosome starting from
chromosome at the cross-point and forms the routing.
hops and less energy usage, and in
generation is repeated, the child route picks the optimal interval of the
eventually has less routing costs and higher expectations than the
and routing is then completed.
4. EXPERIMENTAL RESULT
4.1. Experimental environment
4.2. Assumptions
The transmission and reception costs are set to 16.25
calculated cost of voting was set to 15
nodes, which can select 32 distinct
Figure 4. Route comparison.
original routing and modified routing using proposed scheme. R
points. These are routing information for Parent # 1 and
Thus, the child chromosome starting from Route # 1 changes the path of the parent
point and forms the routing. In Section A, Parent #1 routing has fewer
d less energy usage, and in Section B, Parent # 2 routing is more efficient.
generation is repeated, the child route picks the optimal interval of the parent appropriately and
eventually has less routing costs and higher expectations than the parent generation in this way
ESULT
Experimental environment
Table1. Experiment parameters
The transmission and reception costs are set to 16.25µj and 12.5µj, respectively, and the
calculated cost of voting was set to 15µj. In our experiment, Micaz motes were used as sensor
distinct output levels [9]. Experimental results show that the proposed
e 2018
19
Route # 1 and
and Parent # 2,
oute # 1 changes the path of the parent
1 routing has fewer
2 routing is more efficient. As the
parent appropriately and
parent generation in this way
j, respectively, and the
were used as sensor
Experimental results show that the proposed

20
method does not adjust the maximum output level as the conventional method does, but instead
the proposed method adjusts the maximum output level. Since the report size of the proposed
PVFS and the proposed protocol are the same, the transmission failure is not based on the size of
the report to be transmitted. To increase the reliability and realism of the experiment, experiments
were conducted based on the maximum transmittable distance table according to the node output
level. In the experiment of the proposed method, the list size is 70% of the maximum transfer
range and the number of households is set to not exceed 50 generations. Rather than focusing on
reducing routing costs, we focused on speeding up the routing updates. If the cost is less than the
existing routing, the method is applied to the network immediately without entering the next
generation so the routing can be formed more promptly.
4.3. Experimental result
Figure 5. Number of routing formation successes.
Figure 5 shows the number of routing formation successes according to the number of nodes
placed in the field. Overall, the difference in successes between the existing routing scheme and
the proposed routing scheme is approximately 25%. It is possible to confirm the consistency of
the routing formation success rate, irrespective of the number of nodes. In the conventional
scheme, regardless of the number of nodes, the node with the smallest ID in one cluster is
selected as CH, and thus does not affect the routing formation. In the case of the proposed scheme,
even if the number of nodes is reduced, the performance is constant because the number of nodes
is still high.
0
20
40
60
80
100
120
140
160
180
2000 1800 1500 1200 1000
Num
berofrouting
form
ation
success
Number of node
PVFS Proposed Scheme

21
Figure 6. Routing costs depending on the number of nodes on the field.
Figure 6 shows a comparison of the routing costs consumed by each protocol depending on the
number of nodes on the field. Comparisons were made only among successful routing cases. In
the conventional scheme, energy consumption is independent of the number of nodes, and the
number of nodes and routing cost are not related. In the case of the proposed method, as the
number of generations of GA increases, the energy consumption gradually increases. However, in
the experiment, the creation of the child chromosome for finding a better route path occurs
immediately when the energy consumption is lower than that of the conventional method. In the
energy consumption comparison experiment, the comparison with existing PVFS without the
output level difference application was performed. Moderating the output of the node shows
twice as much energy savings. In existing and proposed schemes with different output levels, the
overall cost difference is approximately 20% due to the location correction of the event detection
CH and the optimization of the routing cost optimization.
0
50
100
150
200
250
300
2000 1800 1500 1200 1000
Routing
cost(μj)
Number of nodes
PVFS PVFS2 Proposed Scheme

22
Figure 7. Number of routing successes according to the cluster size.
Figure 7 shows the number of routing successes according to the cluster size. When the cluster
size is relatively small, there is not a large difference in the number of routing successes between
the two schemes. However, as the size of the cluster increases, the number of routing successes
increases. The number of successes according to the size of the existing scheme was greatly
reduced because a node that does not reach the transmission range frequently occurs due to the
randomly set CH. Alternatively, the proposed method does not have a large drop in selectivity of
the forwarding node in the entire cluster, not just the CH. Additionally, because the algorithm is
focused on immediate routing rather than routing costs, it is expected to be less if the system is set
for cost reduction purposes.
Figure 8. Difference in routing costs according to cluster size.
100
150
200
250
300
350
400
40 43 45 47 50 53 55 57 60
Num
berofsuccess
routing
Size of clusters
PVFS Proposed Scheme
0
50
100
150
200
250
300
350
40 43 45 47 50 53 55 57 60
Routing
cost(μj)
Size of clusters
PVFS PVFS2 Proposed Scheme

23
Figure 8 shows the difference in routing costs according to cluster size. This is similar to the
figure of the routing cost according to the number of nodes, but the cluster size affects the entire
routing success failure and can be interpreted as independent of the routing cost. Routing
optimization of each node was completed without exceeding 10 generations. Routing of all CHs
took less than 1 minute based on the field where more than 2000 nodes were dispatched.
5. CONCLUSIONS
Conventional cluster-based protocols are characterized by routing proceeding through CHs only,
and the dependency on the usability of the CHs is so high that if the CHs are compromised, the
security and energy efficiency of the entire routing using the nodes are damaged. Additionally,
the higher energy usage of CH is expected when it is close to the BS. In this case, faster battery
depletion occurs within those CHs, and the network users have to reroute frequently. In addition,
there is also unnecessary energy consumption, which does not reflect the distance between nodes,
but makes all transmissions the maximum output level. To solve these problems, the proposed
scheme selects CH based on regional suitability, and selects proper delivery nodes considering
every node in the network field. Cost optimization of routing was performed using the GA to
optimize hop count and communication distance. With a proposed scheme, optimization of the
routing cost can be confirmed in the situation where the size of the cluster changes or the number
of nodes in the field changes.
ACKNOWLEDGEMENTS
This research was supported by the MISP (Ministry of Science, ICT & Future Planning), Korea,
under the National Program for Excellence in SW) (2015-0-00914) supervised by the IITP
(Institute for Information & communications Technology Promotion) (2015-0-00914).
REFERENCES
[1] Al-Karaki, Jamal N., and Ahmed E. Kamal. "Routing techniques in wireless sensor networks: a
survey." IEEE wireless communications 11.6 (2004): 6-28.
[2] Perrig, Adrian, John Stankovic, and David Wagner. "Security in wireless sensor networks."
Communications of the ACM 47.6.
[3] Li, Feng, and Jie Wu. "A probabilistic voting-based filtering scheme in wireless sensor networks."
Proceedings of the 2006 international conference on Wireless communications and mobile computing.
ACM, 2006
[4] Zhu, Sencun, et al. "An interleaved hop-by-hop authentication scheme for filtering of injected false
data in sensor networks." Security and privacy, 2004. Proceedings. 2004 IEEE symposium on. IEEE,
2004
[5] Yang, Hao, and Songwu Lu. "Commutative cipher based en-route filtering in wireless sensor
networks." Vehicular Technology Conference, 2004. VTC2004-Fall. 2004 IEEE 60th. Vol. 2. IEEE,
2004.
[6] Yu, Zhen, and Yong Guan. "A dynamic en-route scheme for filtering false data injection in wireless
sensor networks." Proceedings of the 3rd international conference on Embedded networked sensor
systems. ACM, 2005
[7] Geetu, Sonia Juneja. "Performance analysis of SPIN and LEACH routing protocol in
WSN."International Journal of Computational Engineering Research2.5 (2012): 1179-1185.

24
[8] Royer, Elizabeth M., and Chai-Keong Toh. "A review of current routing protocols for ad hoc mobile
wireless networks." IEEE personal communications 6.2 (1999): 46-55
[9] https://www.scribd.com/document/91069327/Micaz-Datasheet-t
[10] D. E. Golberg, "Genetic algorithms in search, optimization, and machine learning," Addion Wesley,
vol. 1989, 1989.
[11] Deb, Kalyanmoy, et al. "A fast and elitist multiobjective genetic algorithm: NSGA-II." IEEE
transactions on evolutionary computation 6.2 (2002): 182-197.
[12] Whitley, Darrell. "A genetic algorithm tutorial." Statistics and computing 4.2 (1994): 65-85.
[13] Benkic, Karl, et al. "Using RSSI value for distance estimation in wireless sensor networks based on
ZigBee." Systems, signals and image processing, 2008. IWSSIP 2008. 15th international conference
on. IEEE, 2008.
[14] Buchli, Bernhard, Felix Sutton, and Jan Beutel. "GPS-equipped wireless sensor network node for
high-accuracy positioning applications." European Conference on Wireless Sensor Networks.
Springer, Berlin, Heidelberg, 2012.
[15] Seon-Phil Kim, Namgi Kim. "Measurement and analysis of the data transmission range using the
transmission power of MICAz." Proceedings of KIIT Summer Conference. 2010
[16 Lin, Shan, et al. "ATPC: adaptive transmission power control for wireless sensor networks."
Proceedings of the 4th international conference on Embedded networked sensor systems. ACM, 2006
[17] Kyoung-Min Park, et al. "Path-Loss Measurements in Indoor Floor Environment at 2.4 GHz " The
Journal of Korean Institute of Communications and Information Sciences 42.8 (2017): 1662-1667.
[18] Mallinson, Michael, Patrick Drane, and Sajid Hussain. "Discrete radio power level consumption
model in wireless sensor networks." Mobile Adhoc and Sensor Systems, 2007. MASS 2007. IEEE
International Conference on. IEEE, 2007
[19 Kubisch, Martin, et al. "Distributed algorithms for transmission power control in wireless sensor
networks." Wireless Communications and Networking, 2003. WCNC 2003. 2003 IEEE. Vol. 1. IEEE,
2003
[20] Lin, Shan, et al. "ATPC: adaptive transmission power control for wireless sensor networks."
Proceedings of the 4th international conference on Embedded networked sensor systems. ACM, 2006
[21] Ogundile, Olayinka O., and Attahiru S. Alfa. "A survey on an energy-efficient and energy-balanced
routing protocol for wireless sensor networks." Sensors 17.5 (2017): 1084.
[22] Yadav, Anju, Vipin Pal, and Ketan Jha. "A Chaotic Genetic Algorithm for Wireless Sensor
Networks." Proceedings of First International Conference on Smart System, Innovations and
Computing. Springer, Singapore, 2018.
[23] Srideviponmalar, P., V. Jawahar Senthil Kumar, and R. Harikrishnan. "Hybrid Genetic Algorithm–
Differential Evolution Approach for Localization in WSN." Intelligent Engineering Informatics.
Springer, Singapore, 2018. 263-271.

25
Authors
Sanghyeok Lim Received a B.S. degree in Digital Information Engineering from
Hanguk University of Foreign Studies in 2017, and is now working toward an M.S.
degree in the Department of Electrical and Computer Engineering at Sungkyunkwan
University.
Taeho Cho Received a Ph.D. degree in Electrical and Computer Engineering from
the University of Arizona, USA, in 1993, and B.S. and M.S. degrees in Electrical
and Computer Engineering from Sungkyunkwan University, Republic of Korea, and
the University of Alabama, USA, respectively. He is currently a Professor in the
College of Information and Communication Engineering, Sungkyunkwan
University, Korea.

GEOGRAPHIC INFORMATION-BASED ROUTING OPTIMIZATION USING GA FOR CLUSTER-BASED WSNS

More Related Content

What's hot (14)

Similar to GEOGRAPHIC INFORMATION-BASED ROUTING OPTIMIZATION USING GA FOR CLUSTER-BASED WSNS (20)

Recently uploaded (20)

GEOGRAPHIC INFORMATION-BASED ROUTING OPTIMIZATION USING GA FOR CLUSTER-BASED WSNS