Voronoi diagram with fuzzy number and sensor data in an indoor navigation for emergency situation

TELKOMNIKA Telecommunication, Computing, Electronics and Control
Vol. 18, No. 4, August 2020, pp. 1990~1997
ISSN: 1693-6930, accredited First Grade by Kemenristekdikti, Decree No: 21/E/KPT/2018
DOI: 10.12928/TELKOMNIKA.v18i4.14905  1990
Journal homepage: http://journal.uad.ac.id/index.php/TELKOMNIKA
Voronoi diagram with fuzzy number and sensor data
in an indoor navigation for emergency situation
Nanna Suryana1
, Fandy Setyo Utomo2
, Mohd Fairuz Iskandar Othman3
, Mohd Nazrin Muhammad4
1,2,3
Fakulti Teknologi Maklumat dan Komunikasi, Universiti Teknikal Malaysia Melaka, Malaysia
2
Department of Informatics, Faculty of Computer Science, Universitas AMIKOM Purwokerto, Indonesia
4
Fakulti Kejuruteraan Pembuatan, Universiti Teknikal Malaysia Melaka, Malaysia
Article Info ABSTRACT
Article history:
Received Aug 6, 2019
Revised Mar 29, 2020
Accepted Apr 13, 2020
Finding shortest and safest path during emergency situation is critical. In this
paper, an indoor navigation during an emergency time is investigated using
the combination of Voronoi Diagram and fuzzy number. The challenge in
indoor navigation is to analyses the network when the shortest path algorithm
does not work as always expected. There are some existing methods to
generate the network model. First, this paper will discuss the feasibility and
accuracy of each method when it is implemented on building environment.
Next, this paper will discuss selected algorithms that determine the selection
of the best route during an emergency situation. The algorithm has to make
sure that the selected route is the shortest and the safest route to the destination.
During a disaster, there are many uncertainties to deal with in determining
the shortest and safest route. Fuzzy logic can be hardly called for to deal with
these uncertainties. Based on sensor data, this paper will also discuss how to
solve shortest path problem using a fuzzy number.
Keywords:
Disaster management
Emergency navigation
GIS
Indoor navigation
Shortest path algorithm
This is an open access article under the CC BY-SA license.
Corresponding Author:
Nanna Suryana,
Fakulti Teknologi Maklumat dan Komunikasi,
Universiti Teknikal Malaysia Melaka,
Melaka, Malaysia.
Email: nsuryana@utem.edu.my
1. INTRODUCTION
Offices and public buildings in many ways are vulnerable to disasters. Fire and earthquake may
happen anytime without prior notice. In such a situation, excellent support in decision-making is of critical
importance to react accurately, fast and effectively [1] in finding the safe exits. The ultimate purpose of
the support system is to avoid any casualties, either the inhabitants or the rescuers. For instance, in case of
a fire, some people may be trapped and have difficulties in finding emergency exits to a safe place. This
scenario becomes apparent that there is a critical necessity to find a solution to this challenging task and
an indoor navigation model that provides routes to safe paths is the best solution.
There are two main research areas when working on indoor navigation; tools and methods.
The former concentrates on the hardware such as facilities and sensors to provide position information in
the disaster areas. The facilities include the signages and special landmarks that could help the victims to find
the directions to safe exits through observation. Meanwhile, navigation sensors such as radar [2-4], inertial
measurement units [5] or wireless positioning technologies [6] provide details information to the victims or
rescuers about their current positions with respect to defined reference in the building. The latter focuses on
the navigation models which study the optimal waypoints to the nearest or the fastest ways to the safe exits.

TELKOMNIKA Telecommun Comput El Control 
Voronoi diagram with fuzzy number and sensor data in an indoor navigation… (Nanna Suryana)
1991
This work normally integrates intelligent systems such as Bayesian filter [7-9] or fuzzy to obtain the best
solution for specific conditions.
In this paper, an indoor navigation model to find the safe exits has been proposed based on
the network model known as Voronoi diagram with fuzzy number. The existing network models have been
studied and the comparison showed the Voronoi Diagram was best suited for the emergency situations. Then,
the most suitable shortest path algorithm was picked. Since there is a lack of literature which discussed indoor
navigation for an emergency with the network model and the shortest path algorithm, the aim of this study is
to build and test indoor navigation for an emergency with the Voronoi diagram and the shortest path algorithm.
Our proposed combination of network model and the shortest path algorithm has achieved the intended goals,
that to find the shortest and safest routes during emergency evacuation. The resident is supposed to obtain
appropriate information which could lead them from one area into other spots up to they attain the refer secure
place through this model. The route direction idea on indoor navigation shall examine several elements [10]
including a particular building map database model, the human redeployment movement and any typical
requirement to a user.
2. 3D GRAPHICAL MODELS
Design plan inherited by mainly the map databases build upon 2D graphical visualization [10]. X3D
for 3D visualisation which is 3D graphics standard or its predecessor virtual reality modelling language
(VRML) is another way to represent the map databases [11]. Yet this model has a shortcoming which it could
not specify the semantics of a given building. Solid learning of map objects is needed than only their topological
relationship to propose a navigation view of a building [10]. CityGML, industry foundation classes, and green
building XML models may resolve this problem. Explained by [12], these models combination may be used to
represent the building fabric into deeper model, which include a geometrical model and a logical model.
Extraction of a clear path to guide the route is hard if only applying a geometrical model. Consequently, since
the logical model may be projected with ease, then this model is selected to use from a route node to another
node [12]. Likewise, the map depiction shall be in the 2-dimension model to produce route guidance shortly. To
make a decision in exigence circumstances is not adequate if only employed the building spatial information. Several
supplementary deliberations shall be highlighted in the process to assist well decision-making.
3. NETWORK ANALYSIS
To choose the optimum route for emergency circumstances on indoor navigation in this study, we
proposed a network design and developed although investigate and simulate existing shortest path algorithm.
Three network models, i.e., adjusting line algorithm [13], Voronoi diagram [14], and Quadtree [15, 16] have
been analysed as input for shortest path algorithm, combined with link cost, to generate evacuation guidance.
Based on available network models, we have noticed, on the one hand, Quadtree has a drawback i.e. fall
through in velocity consequence of peak complexity. However, it has good accuracy since this model is highly
specified. Furthermore, "move into corridor" is adopted by adjusting line algorithm and Voronoi diagram is
simpler and faster than Quadtree. Nevertheless, between all the networks, adjusting line algorithm has the worst
network accuracy than the other. Hence, based on speed and accuracy, the Voronoi diagram is the prime
selection for the network model. This diagram represents a subdivision of space into regions whose points are
closer to a generating vertex than any other element. Centerline algorithms using the Voronoi diagram begin
by sampling the polygon boundary and constructing the Voronoi diagram. The intersections between
the Voronoi edges converge to the polygon centerline, as the boundary sampling rate is increased. One problem
with this method is the difficulty in joining centerline segments from separate but adjacent hallway polygons.
4. SHORTEST PATH ALGORITHM
Having chosen the network diagram, the next logical step is to choose the best shortest path algorithm
for this indoor navigation. Five algorithms, i.e., Dijkstra [17], Floyd-Warshall [18-20], A* Search [21],
Bellman-Ford [22], and Johnson’s [23] have been taken into consideration in this project. A* Search is the best
shortest path algorithm which could provide the top accuracy. It works mechanism based on Euclidean distance
to the destination node sans forecasting the entire network. Hence, there possibly some cases whereupon
the optimum route would be displaced. Furthermore, Floyd-Warshal algorithm is weak in consequence of n3
complexity. To calculate the optimum route, Dijkstra, Bellman-Ford and Johnson’s algorithm might be great
preferences. Nevertheless, the Dijkstra algorithm has a better achievement than the others. We don't employ
negative weights from Dijkstra in our study since it might collapse dealing on negative link cost.

 ISSN: 1693-6930
TELKOMNIKA Telecommun Comput El Control, Vol. 18, No. 4, August 2020: 1990 - 1997
1992
5. RESEARCH MODELS
Figure 1 shows the research model on the safest and shortest route. The network model is produced
by network analysis from a 2D building environment transformation. For producing the evacuation directive,
on the shortest path algorithm, this network model conjoined with a link cost as an input. Based on
the building form through the emergency period which is described with a fuzzy crisp number, the link cost is
seriously influenced by this condition. This part of the paper discusses how the shortest and the safest route
is defined. The main consideration of route guidance during the emergency time is speed and accuracy.
In term of speed, the route guidance must be generated fast and very adaptive to network condition. Next,
the shortest route is not always the safest route during emergency time. Therefore, in term of accuracy,
the generated route must avoid dangerous places. Each edge of the network is represented by membership
function that considers many factors during the disaster. It is generalised based on the canonical representation
of operations on fuzzy triangular numbers to define the shortest route. This method is more efficient because
the summing operation and the ranking of fuzzy numbers can be done in an easy and straight manner.
Speed and accuracy are parameters that may be involved in determining the calculation of safest path.
The selection is based on the reuses of the existing algorithm and as well as based on technical analysis and
economic consideration.
Figure 1. Effects of selecting different switching under dynamic condition
6. RESULTS
The following presents the results of our analysis and design which started from pre-processing
building map, generating building network, and generating shortest path. This section is intended to provide
insight into the whole processes of generating, facilitating and building the possible network topology including
the possible network path. From this, it allows to develop a network topology table in which all different
possible route or path analysis can be performed.
6.1. Pre-processing building map
For testing, we had used map of the ground floor of Zone A and B, Faculty of Information and
Communication Technology, Universiti Teknikal Malaysia Melaka. The map was provided by Pejabat
Pembangunan Universiti Teknikal Malaysia Melaka. The initial format of the map is in AutoCAD data. We had
chosen PostgreSQL/PostGIS to store the data. To Test the data, we had used QGIS Desktop because it can store and
retrieve data from PostgreSQL/PostGIS and used spatial analysis tools from PostGIS combined with pgRouting.
6.2. Generating building network
The building network can be generated by performing some steps. From the base map, we had created
a polygon which represents rooms and accessible areas of the building as shown in Figure 2. After creating
the polygon, we created a Voronoi diagram using ST_StraightSkeleton function from PostGIS.
The result can be seen in Figure 3. Next, we created network path by cleaning up lines which have no intersection
in one of their vertex (dead end) as shown in Figure 4. Lastly, we created a network topology in
a database using the pgr_createTopology function from pgRouting. Network topology’s table is shown in
Figure 5. The cost and multiplier column will define the link cost. In the event of disaster, very unlikely that
the victims have sufficient time to response in appropriate manners. Many cases the victims will not be able to
assess their situations properly, in order to minimize the damage from the disaster. For examples, in the event of
fire, victims should feel the doors with the back of hand before opening them, which if they are warm, then

1993
the fire is on the other side. In the actual situation, victims are rushed to find the safe exits, thereby there are
possibilities that they endangered themselves by entering the fire area unintentionally. In this regard, there is
a need to use sensors to sense the disaster’s surroundings to manage the risk of injury or casualty.
Figure 2. Building polygon Figure 3. Voronoi diagram
Figure 4. Network path
Figure 5. Network topology table
The use of suitable sensors will provide a better evacuation guidance. In an indoor environment,
among possible disasters to dealt with are fire and explosion. In explosion, there is a high chance that it will
cause a fire as well. Therefore, in both events, smoke and heat sensors can be installed to check
the surrounding. In a compact version, both sensors can be enclosed together as a single unit, called as a node,
with wireless connectivity such as Wi-Fi, Bluetooth or IEEE 802.15.4 standard. In this work, IEEE 802.15.4
is preferred because of the low power consumption. These nodes can be connected in a mesh network with

 ISSN: 1693-6930
1994
a gateway to the internet. Each node will check the presence of smoke and measure the surrounding temperature
(in Celsius), in 30 seconds interval. Every interval elapsed, the data will propagate until it reaches the gateway,
which will send the data to the cloud. The data payload is 1 byte as shown in Figure 6.
Figure 6. Data payload
For instance, at ambient temperature (27C) without presence of smoke, the payload will be 27h
(0010 0111) and if smoke was presence, and temperature rise to 45C, the payload will be C5h (1100 0101).
The timestamp will take 8 bytes and another 1 byte is for the location ID. In overall, the payload has a size of
10 bytes. The data propagates from the source to the destination, gateway, through Ad Hoc on-demand distance
vector (AODV) routing protocol. One of the advantages in using AODV is it can respond very quickly to
the topological changes that affect the active routes. This is crucial especially if the nodes happened to damage
during the disaster, the communication between other nodes to the gateway will not be interrupted.
The data that reached the gateway will be stored in the global server such as cloud. The server will
detect any anomaly and send emergency notification to the affected users. At the same time, the server will
request the sensor data more frequent from the nodes, in every 10 seconds. This is another practical justification
to adopt the AODV which the main advantage of this protocol is that routes are established on demand.
The emergency notification contains information of level of disaster and location of disaster. In the Figure 1,
the building equips with a number of nodes that are connected in mesh. Every node will report to the gateway
at the specified interval and the gateway sends the information to the server. In case of fire, in the proximity of
node 1, as illustrates in the figure, the server detects the disaster by observing a rise in temperature. The server
sends emergency notification to the users in the building to leave and avoid that particular area. The presence
of smoke and high heat will give a red alert and can represent as constraint value in order to determine
the shortest and safe path based on Dijkstra algorithm.
6.3. Generating shortest path
 Shortest path without constraint
We use Dijkstra algorithm to find the shortest path to the exit. Dijkstra shortest path is performed by
calling the PGR Dijkstra function from pgRouting. The parameter for this function is source node, target node
and cost. Link cost is calculated by multiplying cost and multiplier field. The result is presented in Figure 7.
Figure 7. Fire at an established disaster management setup

1995
 Shortest path with constraint
To set constraint, we set multiplier field on the database. To define the hazard area, the multiplier
value is set to 99999, as shown in Figure 8. Next, we performed the Dijkstra shortest path to select the best exit
as we did without constraint. The result can be seen in Figure 9.
Each node will check the presence of smoke and measure the surrounding temperature (in Celsius),
in 30 seconds interval. Every interval elapsed, the data will propagate until it reaches the gateway, which will
send the data to the cloud. The data payload is 1 byte as shown in Figure 10. The emergency notification
contains information of level of disaster and location of disaster. In the Figure 7, the building equips with
a number of nodes that are connected in mesh. Every node will report to the gateway at the specified interval
and the gateway sends the information to the server. In case of fire, in the proximity of node 1, as illustrates in
the figure, the server detects the disaster by observing a rise in temperature. The server sends emergency
notification to the users in the building to leave and avoid that particular area as shown in Figure 11.
The presence of smoke and high heat will give a red alert and can represent as constraint value in order to
determine the shortest and safe path based on Dijkstra algorithm.
Figure 8. Set multiplier to define hazard
Figure 9. Shortest path with constraint
Figure 10. Data payload which send to the cloud

 ISSN: 1693-6930
1996
Figure 11. Emergency notification
7. DISCUSSION
From sub-section 3 and 4, it becomes apparent that the existing model and algorithm have pros and
cons to produce the shortest and safest route. Based on this strength and weakness analysis, we propose
a hybrid shortest and safest route algorithm which is a combination of Voronoi diagram and Dijkstra algorithm
using a fuzzy number. The proposed indoor navigation model had successfully provided the shortest path to
the safe place while considered the hazardous area such as fire. Currently the constraint that set as the hazard
was fixed in the simulations. However, the value can be dynamically changed if more information from
the surroundings were available. The potential hazards such as heat intensity can be sensed through temperature
sensor that could provide spectrum of dangerous zones; low heat, medium heat and high heat. The humidity
sensor also can be used to check the intensity of water vapour in particular areas. In the fire case, high humidity
areas would engulf in flames slower compared to dry places. Therefore, with reference to the temperature and
humidity spectrums, the algorithm could accommodate less risk situations when computing the shortest as well
as the safest routes to exit. Apart than that, the model could integrate with navigation sensors to exactly
determine one’s position with respect to the building. This is imperative as wrong position would lead to fatal
decision when determining the shortest path. One of the best options for positioning and tracking the victims
is by using device free localization (DFL) technique [24]. This technique does not require the user to wear any
tracking device on the body, which is essential in time-critical situations. One of DFL methods that could be
applied is known as radio tomography imaging (RTI) [25]. The RTI system will determine the signal
disturbance detected during radio transmission between the wireless nodes that are fixed along the corridors as
the potential human presence. Through this model, occupants are expected to get accurate information that can
guide them from one room to another until they reach designated place. More importantly, the route generated
must be not only be short but also safe.
8. CONCLUSION
An indoor navigation model using sensor data has been proposed and aimed to standardise and
investigate methods and algorithms for navigation in buildings. An inclusion of Voronoi diagram combined
with a classic Dijkstra algorithm gives a significant early result especially in term of speed and accuracy.
The combination is considered to be complementary to each other to produce a better result. Fuzzy numbers
can be used to deal with the uncertain environment. However, this hypothetical conclusion needs to be validated
with a set of experiment and testing. The criteria for generating membership function of the fuzzy set must be
studied as well in future works.
REFERENCES
[1] Van Borkulo, Erik, et al., "Services for an emergency response system in the Netherlands," Second Symposium on
Gi4DM, 2006.
[2] M. Moallem, K. Sarabandi, "Polarimetric study of MMW imaging radars for indoor navigation and mapping," IEEE
transactions on antennas and propagation, vol. 62, no. 1, pp. 500-504, 2013.
[3] Sulistyaningsih Y. P., Saputera M., Wahab Y., Y. Maulana, "Design of radar display of Indonesian airspace
monitoring application," TELKOMNIKA Telecommunication Computing Electronics and Control, vol. 17, no. 3,
pp. 1176-1184, 2019.
[4] M. F. E. Purnomo, et al. "Development of triangular array eight patches antennas for circularly-polarized synthetic
aperture radar sensor," TELKOMNIKA Telecommunication Computing Electronics and Control, vol. 18, no. 2,
pp. 631-639, 2020.

1997
[5] M. N. Muhammad, et al., "Indoor pedestrian tracking using consumer-grade inertial sensors with PZTD heading
correction," IEEE Sensors Journal, vol. 18, no. 12, pp. 5164-5172, 2018.
[6] X. Tian, et al., "Optimization of fingerprints reporting strategy for WLAN indoor localization," IEEE Transactions
on Mobile Computing, vol. 17, no. 2, pp. 390-403, 2017.
[7] W. Wang, Wei-Shinn Ku, "Dynamic indoor navigation with bayesian filters," SIGSPATIAL Special, vol. 8, no. 3,
pp. 9-10, 2017.
[8] O. Kaltiokallio, R. Hostettler, N. Patwari, "A Novel Bayesian Filter for RSS-based Device-free Localization and
Tracking," IEEE Transactions on Mobile Computing, pp. 1-16, 2019.
[9] J. Hwang, et al., "Development of error reduction model using Bayesian filter for AUV navigating under moving
ice," IEEE/OES Autonomous Underwater Vehicle Workshop (AUV), pp. 1-6, 2018.
[10] Pierre-Yves Gilliéron, et al., "Indoor navigation performance analysis," Proceedings of the 8th European navigation
conference GNSS. No. CONF, pp. 1-9, 2004.
[11] U. Rueppel, K. M. Stuebbe., "BIM-based indoor-emergency-navigation-system for complex buildings," Tsinghua
science and technology, vol. 13, no. S1, pp. 362-367, 2008.
[12] S. Pu, S. Zlatanova., "Evacuation route calculation of inner buildings," Geo-information for disaster management,
Springer, Berlin, Heidelberg, pp. 1143-1161, 2005.
[13] F. Lyardet, J. Grimmer, M. Muhlhauser, "CoINS: Context sensitive indoor navigation system," Eighth IEEE
International Symposium on Multimedia (ISM'06), pp. 209-218, 2006.
[14] J. Liu, et al., "Semi-automated processing and routing within indoor structures for emergency response applications,"
Cyber Security, Situation Management, and Impact Assessment II; and Visual Analytics for Homeland Defense and
Security II, vol. 7709, pp. 77090Z, 2010.
[15] H. Samet, "The quadtree and related hierarchical data structures," ACM Computing Surveys, vol. 16, no. 2,
pp. 187-260, 1984.
[16] M. Meijers, S. Zlatanova, N. Pfeifer, "3D geoinformation indoors: structuring for evacuation," Proceedings of Next
generation 3D city models, vol. 6, p. 11-16, 2005.
[17] E. W. Dijkstra, "A note on two problems in connexion with graphs," Numerische mathematik, vol. 1, no. 1,
pp. 269-271, 1959.
[18] R. W. Floyd, "Algorithm 97: shortest path," Communications of the ACM, vol. 5, no. 6, 1962.
[19] S. Warshall, "A theorem on boolean matrices," Journal of the ACM, vol. 9, no. 1, pp. 11-12, 1962.
[20] K. Gutenschwager, et al., "The shortest path: Comparison of different approaches and implementations for
the automatic routing of vehicles," Proceedings of the 2012 Winter Simulation Conference (WSC) IEEE,
pp. 1-12, 2012.
[21] P. E. Hart, N. J. Nilsson, B. Raphael, "A formal basis for the heuristic determination of minimum cost paths," IEEE
transactions on Systems Science and Cybernetics, vol. 4, no. 2, pp. 100-107, 1968.
[22] Y. Huang, H. Sheng, J. Chen, "Intelligent Congestion Avoidance Algorithm and System-Application of Data
Vitalization," 14th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing, pp. 847-856, 2014.
[23] D. B. Johnson, "Efficient algorithms for shortest paths in sparse networks," Journal of the ACM, vol. 24, no. 1,
pp. 1-13, 1977.
[24] W. Ruan, et al., "Device-free indoor localization and tracking through human-object interactions," IEEE 17th
international symposium on a world of wireless, mobile and multimedia networks (WoWMoM), pp. 1-9, 2016.
[25] S. Denis, et al., "Multi-frequency sub-1 GHz radio tomographic imaging in a complex indoor environment,"
International Conference on Indoor Positioning and Indoor Navigation (IPIN), pp. 1-8, 2017.

Voronoi diagram with fuzzy number and sensor data in an indoor navigation for emergency situation

More Related Content

What's hot (17)

Similar to Voronoi diagram with fuzzy number and sensor data in an indoor navigation for emergency situation (20)

More from TELKOMNIKA JOURNAL (20)

Recently uploaded (20)

Voronoi diagram with fuzzy number and sensor data in an indoor navigation for emergency situation