1

Performance evaluation of a Volcano MonitoringSystem Using Wireless Sensor Networks

Roman Lara-Cueva, Member, IEEE, Antonio Caamano, Member, IEEE, Marco Zennaro, Member, IEEEand Jose Luis Rojo-Alvarez, Senior Member, IEEE

AbstractWireless Sensor Network have become critical inthe evolution of Telecommunications. Our interest is related tomonitor an Active Volcano using wireless sensor networks wherethe requirement of real-time is mandatory, due to the necessityof access immediately of the signals derived from a naturaldisaster in order to determine emergency early warnings. Ouraim was to determine the number of sensors will be deployedin a Volcano Monitoring System based on simulation results andcorroborated with an in-situ testbed. We used ns-2 as simulationtool, where two scenarios were evaluated. This study determinedthe optimal scenario in volcano monitoring is Randomly withmaximum eighteen nodes to satisfy metrics as throughput, timedelay and packet loss. We deployed sixteen sensors in a strategyarea at Cotopaxi Volcano, the information was obtained duringthree days of continuous monitoring. This information was sentto a surveillance laboratory located 45 km away from the stationplaced at the volcano, a WiFi-based long distance technology wasused to this purpose. Volcanic information was processed in time,frequency and scale domain, a spectral pattern of seismic eventsdetermined four kinds of events, corresponding to a long periodevents, volcano tectonic earthquakes, volcanic tremor, and hybridevents.

Index TermsWSN, 802.15.4, throughput, delay, packet loss,monitoring system, volcano applications.

I. INTRODUCTION TO WSN

Wireless Sensor Network (WSN) have become critical inthe evolution of Telecommunications, its constant evolutionpermits the possibility to implement devices with low cost andenergy autonomy, without periodic maintenance, to be capableof obtaining environmental information, and for these reasonsCyber-Physical Systems (CPS), Internet of Things (IoT), andSmart Cities are new research topics based on technologieslike WSN, all of them are based in a similar infrastructure ofevery heterogeneous networks where data must be transmitted,processed, and finally enable people through any applicationto monitor or to control objects [1][6].

Ecuador has a special interest to use WSN in volcanomonitoring applications, as it is located in the Pacific Ringof Fire a place with high seismic activity. WSN systems are

Manuscript received July 5, 2014R.A. Lara is with Wireless Communication Research Group (WiCOM)

and Ad Hoc Networks Research Center (CIRAD) of the Depart-ment of Electrical and Electronic Engineering, Universidad de lasFuerzas Armadas ESPE, Sangolqu-Ecuador, 171-5-231B e-mail: (seehttp://wicom.espe.edu.ec/contactos.html).

A. Caamano and J.L. Rojo are with Information and CommunicationTechnology Department, Rey Juan Carlos University, Camino del Molino s/n28943, Fuenlabrada-Madrid, Espana

M. Zennaro are with ICT for Development Laboratory , The Abdus SalamInternational Centre for Theorical Physics, Trieste-Italy

much cheaper and reliable than bulky and energy-hungry tradi-tional systems. Currently, South America lacks of permanentmonitoring systems deployed in active volcanoes, but someWSN-based systems have been installed just in a couple ofweeks. Volcano monitoring using WSN still requires furtherresearch in order to present information in real-time and tolaunch an early emergency warning. The main constraint tobe considered as real-time systems is the time delay causedby processing data, there are some solutions at Network andMAC layer referred to data acquisition in-situ, data gatheringand data dissemination which significantly reduces time delay,but it is impossible to give an early warning with this kind ofsystems, because data are processed off-line in a far distancedsurveillance laboratory. The time delay related to digital signalprocessing and the signal propagation must be solved withinappropriate processing and telecommunication techniques [7][9].

Our aim was to determine a number of sensors that max-imize the network capacity, we considered as main metricsthroughput, time delay and packet loss in the WSN, andthen corroborated by an in-situ testbed deployed at Cotopaxivolcano [10].

The rest of the paper is organized as follows. Section2 summarizes previous research on the subject. Section 3describes the performance evaluation of the WSN in detail.Sections 4 and 5 describe the simulation results obtainedand the experimental study performed, respectively. Finally,Section 6 presents our conclusions and future work [11].

II. RELATED WORK

Our interest consist in determining the network behavior,for that reason we must study the performance network, thiscan be evaluated by some Quality of Service (QoS) metricssuch as: availability, reliability, response time, time delay,throughput, bandwidth capacity, and packet loss ratio. In orderto offer real-time in WSN with guaranteed QoS metrics, thenetwork must be analyzed in a different way than traditionalreal-time systems, as far as WSN requires to need severechallenges due to its wireless nature, distributed architectureand dynamic network topology. The state of the art of real-time solutions currently developed have been presented withemphasis at level of MAC, routing, data processing, and crosslayer, this denote a direct relationship between real-time andQoS metrics, as well as new general concepts related to real-time WSN systems [12] [13]. Real-Time (RT) WSN could bedefined as a WSN capable of ensuring a Maximum Sustained

2

Traffic Rate (Throughput), a Minimum Latency, and PacketLoss as main QoS metrics. An ideal development processshould start from the theoretical analysis of the protocol toprovide bounds and information about its performance [14],[15], it must be verified and refined by simulations [16][21], and finally confirmed in a testbed [22], [23]. We foundseveral works which presented a mixed analysis, since in realscenarios it is possible to obtain measures of main metrics as:RSSI, PER, and EED through tools developed by manufactures[24][26].

III. PERFORMANCE EVALUATION OF THE MAC PROTOCOL

A. Real-Time WSN for a Volcano Monitoring System Require-ments

For our application we have to consider the environmentpresented by a Volcano a wild terrain and a lack of energyto implement a WSN. A mesh topology would present the bestway to communicate sensors in this kind of scenario, speciallybecause we have to define the position of the nodes accordingto the requirements that an in situ visit could give us accordingto the variables to monitor.

It is developed a research about topologies for ad hocnetworks and it is presented the concept of tessellation, inwhere, this model could be applied in WSN to obtain regularnetwork formed by geometrical figures [27], that can bespecified using the notation of Schlafli [28].

B. Simulation Environment

There is a wide range of simulators that can be used to testWSN in order to obtain several results to be analyzed basedon [29] we have chosen ns-2 as simulation tool.

In order to obtain the performance of the network throughsimulation we have chosen two scenarios: tessellation andrandomly, for the first one we have chosen a triangulartessellation pattern network {3, 6}, in the second one wedefined a random position of the sensor nodes placed on theplane at a distance of 30 meters each (typical mean value forconnection in practice).

The number of nodes (n) in the triangular tessellation couldbe obtained as function of the number of layers C, this is,

n = 1 + 3C(C + 1). (1)

In both scenarios we started with 6 nodes growing until60 nodes in 6 nodes step each, we defined one coordinator,all transmissions nodes were directed to the coordinator. Weassumed an event occurred with a duration equal to 220seconds in an approximated area of 300 300 m.

In our simulation process it was necessary to determinea number of replications to reduce the mean square error,consequently, we defined for tessellation scenario to run 6replications, and for randomly scenario the number of repli-cations depends of the number of the nodes in the scenario,in other words, we run n replications in each scenario.

The main simulation parameters have been defined amongothers: time simulation, topology, routing protocol, transmis-sion rate, etc., we could cluster all of them in three main

0

20

40

60

80

0

0.02

0.04

0.06

0.08

0.1

0.12

0

50

100

150

200

250

300

Number of nodes (n)Delay (s)

2^(

n+

1)

(%)

Fig. 1. XXXXXX

groups: general parameters, power parameters, and node pa-rameters, the rest of the parameters to simulate the networkmodel are detailed in the Tables I, II and III respectively.

TABLE IGENERAL PARAMETERS

Parameter ValueRadio Propagation Model Two-Ray GroundRouting Protocol AODVRaw Bit Rate (kbps) 250Antenna Type DirectionalSimulation Time (s) 260Simulation Time (s) 220

TABLE IIPOWER PARAMETERS

Parameter ValueTransmission Power (dBm) 0 (1mW)Sensitivity (dBm) -94Transmission antenna gain Gt (dB) 1.0Reception antenna gain Gr (dB) 1.0Trajectory loss (dB) 1.0

TABLE IIINODES PARAMETERS

Parameter ValueTraffic type FTPTraffic direction all to CoordinatorPackage size 55 bytesNumber of Coordinators 1 coordinatorDistance between nodes 30 mNumber of nodes 1 to 45 nodes

EnabledBeacon mode Beacon Order:3

Superframe Order:3

The scenarios of both topologies of the networks defined forour simulations are shown in Fig. 2, to obtain the results ofnetwork performance and verify the operation with an optimalnumber of sensors.

C. Performance Metrics

We selected three main metrics required for a real-timemonitoring: normalized throughput (), end-to-end delay

3

150 100 50 0 50 100 150150

100

50

0

50

100

150Tessellation topology

Position x (m)

Po

sitio

n y

(m

)

Coordinator

FFD Nodes

150 100 50 0 50 100 150150

100

50

0

50

100

150Randomly topology

Position x (m)

Po

sitio

n y

(m

)

Coordinator

FFD Nodes

Fig. 2. Tessellation and Randomly Scenarios evaluated.

(EED), and packet loss (PL). As mentioned in the previousworks there are other metrics that can be considered but wecould determine that all of them have a direct relation to ourmain metrics, for example: duty cycle, energy consumption,average jitter, load factor, and traffic type.

In several meetings maintained with experts in volcanologyfrom Instituto Geofsico de la Escuela Politecnica Nacional(IGEPN), we determined that the system must be able to workin a permanent way; to monitor several variables in a volcano itis not effective to have a WSN in a saving power mode, for thatreason we did not consider power consumption metrics. Forthose reasons, with respect to our metrics we need a maximum, PL must be less than 20%, a minimum EED, and at least5 stations must be necessaries. For we used the informationthat permitted us to define the Eq.(2), and we calculated defined in Eq.(3), meanwhile for EED and PL we used theinformation obtained in the trace file. We use a tool developedby Jaroslaw Malek named tracegraph to analyze the trace fileyielding [30],

Throughput = 8B

ttx

[bits

s

](2)

where, B is the number of transmitted bytes, and ttx is thetime of transmission in seconds. Also, parameter representsthe normalized throughput and is obtained as

=Throughput

RBR(3)

where, RBR is the theoretical bite rate = 250 kbps.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

6 12 18 24 30 36 42 48 54 60 66nodes (n)

Normalized Throughput vs n

a) Tessellation Scenario

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

6 12 18 24 30 36 42 48 54 60 66nodes (n)

Normalized Throughput vs n

b) Randomly Scenario

Fig. 3. Normalized throughput as function of n in both: Tessellation andRandom scenarios, this metric had an indirect relationship related to thenumber of nodes.

IV. SIMULATION RESULTS

Figures 3, 4, and 5 show our main metrics related to thenumber of nodes obtained in Tessellation and Randomly sce-narios. Figure 3a shows an irregular decay of in Tessellationscenario, its maximum and minimum values were 0.58 and0.28 respectively; its boxplots corresponded a line becauseits standard deviation was insignificant. Meanwhile, Fig. 3bshows a directly relationship between and n in Randomlyscenario, its maximum and minimum values were 0.57 and0.41 respectively; its boxplot showed a significant standarddeviation due to its own random nature and the number ofreplications we have done.

The data suggest PL in both scenarios presented an incre-ment as an exponential function of n, in Fig. 4a we determinedfor Tessellation scenario in the range of nodes from 36 to 48presented an irregularity, meanwhile in Fig. 4b we observedRandomly scenario increase its PL directly to n, the maindifference between both scenarios was the Randomly scenariohad less PL than Tessellation scenario.

Finally, Fig. 5 shows the EED, we observed that in Fig.5a Tessellation scenario presented more irregularity than Ran-domly scenario related to this metric, both scenarios presenteda mean value around to 3 ms.

We found based on the requirements gave from IGEPNrespect to PL must be less than 20%, Tessellation reached this

4

0

50

100

150

200

250

300

350

400

450

500

6 12 18 24 30 36 42 48 54 60 66nodes (n)

Pa

ke

t L

oss

PL vs n

a) Tessellation Scenario

0

50

100

150

200

250

300

350

400

450

500

6 12 18 24 30 36 42 48 54 60 66nodes (n)

Pa

ke

t L

oss

PL vs n

b) Randomly Scenario

Fig. 4. Packet Loss as function of n in both: Tessellation and Randomscenarios, this metric had a direct relationship related to the number of nodes.

value when n = 12 and Randomly reached this value when n =18, Randomly scenario permit us to use more nodes with sameor better characteristics of and EED presented in Tessellationscenario.

V. RESULTS FROM COTOPAXI VOLCANO DEPLOYING

The Wireless Communications Research Group (WiCOM)from Universidad de las Fuerzas Armadas ESPE, developed afirst attempt for replicating the predecessors works by usingMicaz and Iris platform. Thus, Cotopaxi, which is currentlythe highest snowcapped volcano on Earth, was selected fordeploying a WSN. This work was deployed at an altitude of4870 meters, where 16 sensor nodes were implemented, anddata were collected continuously for three days, Table. IVsummarize the comparison among previous works and ourdeployment.

0

0.5

1

1.5

2

2.5

3

3.5

4

x 103

6 12 18 24 30 36 42 48 54 60 66nodes (n)

En

d

to

End

EED vs n

a) Tessellation Scenario

0

0.5

1

1.5

2

2.5

3

3.5

4

x 103

6 12 18 24 30 36 42 48 54 60 66nodes (n)

En

d

to

En

d

EED vs n

b) Randomly Scenario

Fig. 5. End-to-end delay as function of n in both: Tessellation and Randomscenarios, this metric had a quasi-constant relationship related to the numberof nodes.

TABLE IVCOMPARISON DEPLOYMENTS

WSN Variables/ Number Operation Frequency DurationDeploy. Platform sensors frequency Sampling (Days)

(MHz) (Hz)[31] Aa/Micaz 3 2400 100 2[32] SAb/Tmote 16 2400 100 19[33] SA/Imote2 5 900 1000 AbidingProposed Sc/MicazWork e Iris 16 2400 100 3

aAcusticbSeismo-AcusticcSeismo

a) MICAz motes deployed on Volcano b) IRIS motes deployed on Volcano

Fig. 6. Wireless Sensor Networks deployed on Cotopaxi Volcano

One of the problems of this network is their dependence on

5

an aggregator, added to the information that must be storedbefore being transmitted to the surveillance laboratory througha wireless link of 40 km of distance where information isprocessed in order to determine whether or not the events arefalse alarms, this is the main problem to be solved, the signalprocessing is off-line, because of that, it is not possible toguarantee a real-time monitoring data [34].

Our first visit to the Cotopaxi Volcano was for locatingthe geographic coordinates for placing the wireless commu-nications system (0o 39 49 S, 78o 26 17 W), and todetermine the necessary requirements for WSN deployment onCotopaxi Volcano. In our second visit WSN were deployed,at an altitude of 4870 meters, two differents WSNs, oneof them consisted of 10 motes MICAz, and another with 6motes IRIS with MTS400 and MTS310 sensor cards usingtwo gateways MIB520 each. Mote Config 2.0 was the softwareused to configure nodes. FTSP was implemented for timesynchronization. Fig. 6 shows the location of the two networksdeployed. Data were collected continuously for three days. Theenergy problem was solved with a generator placed in situ. Theinformation was stored in a central station placed in situ, thenit was transmitted to the surveillance laboratory, located at adistance of 40 km from volcano to ESPE, through a wirelesslink.

A. WiFi-Based Long Distance Link

A wireless link was used for transporting data sensed byWSN, it has proved to be cost effective for long distanceapplications. The two major limitations for using WiFi overlong distances (WiLD) are the requirement for line of sightbetween the endpoints and the vulnerability to interference inthe unlicensed band. Two further hurdles have to be overcomewhen applying WiLD Technology, power budget and timinglimitations. The first was easily solved by using high gaindirectional antennas, while the timing issue was addressed bymodifying the media access mechanism, as done by the TIERgroup at the University of Berkeley [35].

For our purposes IEEE 802.11b was selected, basically,because the 2,4 GHz ISM band presents less losses that5,8 GHz ISM band, we used antennas gain of 24 dBi andthe transmission power of 1 watt. Meanwhile in the MACsublayer, three types of limitations can be extracted: the timerwaiting of ACKs, RTS/CTS and the definition of time relatedto the slottime. We used Alix boards, in which we embeddeda middleware that permit us to modified these parameters inorder to link endpoints according. The performance of the linkwas determined using DITG traffic injector, the injection timewas 1 minute, Fig. 7 is a portion of total file analyzed showingthat the mean throughput obtained is less than 2 Mbps anda packet loss is less than 5%, this data rate is enough fortransmitting information sensed by our WSN, but the packetloss must still be improved.

B. Seismic Signal Analysis

For a better representation of the signal is necessary tohave a representation in the time and frequency domain. UsingFourier Transform, it is possible to detect the presence of a

a) Throughput b) Number of packet loss

Fig. 7. Parameters of Cotopaxi-ESPE link

certain frequency but it does not provide information aboutthe evolution over time of the spectral characteristics, forthis reason it is important to note that a spectrogram wasconstructed using the Short Time Fourier Transform (STFT)with a sampling frequency of 100 Hz.

The information in time and frequency domain of seismicsignals on a determined time often cannot be known, therefore,we cannot define if the spectral component exists at any instantof time, the only thing that can be distinguished are the timeintervals at certain frequency bands where we can find spectralcomponents.

Since seismic signals are non-stationary, it is necessaryto vary the size of the window, for this reason, we usedscale domain through wavelet analysis. The seismic signalsrecorded was processed using discrete and continuous wavelettransform, in order to discriminate the occurrence of a par-ticular seismic event. The mother wavelet employed is DB6(Daubechies 6), in the project were used decomposition levels2 and 4 for the records of MICAz motes and IRIS motesdeployed respectively due to the amount of data they hold.We must define if the event belongs to a volcanic tremor,hybrid, long period or volcano tectonic events that permit usto determine an abnormal behavior.

In Fig. 8, it was determined that the spectral componentspresent in the frequency range [1,25 - 2,5 Hz] are stableand present a low energy, this signal corresponds to a noiseseismic event. Meanwhile, in Fig. 9, the signal spectral contentbetween [2,5 - 5 Hz] shows a seismic noise in a short period,this is due to sudden temperature changes. In Fig. 10, thesignal between [5 - 10 Hz] is equally distributed in energy,its acceleration remains constant during all time, this eventcorrespond to a volcanic tremor, this event is associated togas outlet due to high pressures. Finally Fig. 11 shows anothercase, a high frequency volcanic tremor, signal between [10 -20 Hz] it is related to strong gas outlet inside of the crater.

VI. CONCLUSION AND FUTURE WORKS

We determined the most optimal scenario in volcano mon-itoring is Randomly, it presents the best performance relatedto all our metrics, to reach the PL 20% and the mean valuereached to EED = 3 ms, with its consequent of maximizationthroughput. In this range we determine that the maximumthroughput is approximately equal to 145 kbps. This wascorroborated with an in-situ deployment, we determine a PLequal to 25%, the mean value of EED = 1,1 s and a maximumthroughput equal to 130 kbps, the main difference between

6

Fig. 8. Noise seismic event determined after digital processing signal

Fig. 9. Short period noise seismic determined after digital processing signal

simulation and testbed was EED, data suggest this differencedue to the processing data must take to much time and thisvalues is unconsidered in simulation tools.

As future work we will modeling this system consideringthe value of EED related to processing data, moreover we areinterested in feature extraction of volcano signals in order toclassify automatically these kind of events.

VII. ACKNOWLEDGMENTS

The authors gratefully acknowledge the contribution ofUniversidad de las Fuerzas Armadas ESPE for the economicalsupport in the development of this project through the WirelessCommunications Research Group (WiCOM).

REFERENCES[1] R. Baheti and H. Gill, Cyber-physical systems, The Impact of Control

Technology, pp. 161166, 2011.[2] R. Poovendran, Cyberphysical systems: Close encounters between two

parallel worlds [point of view], Proceedings of the IEEE, vol. 98, no. 8,pp. 13631366, 2010.

[3] N. D. Stojanovic, L. Stojanovic, and R. Stuehmer, Tutorial: personalbig data management in the cyber-physical systems-the role of eventprocessing, in Proceedings of the 7th ACM international conference onDistributed event-based systems. ACM, 2013, pp. 281288.

[4] L. Atzori, A. Iera, and G. Morabito, The internet of things: A survey,Computer Networks, vol. 54, no. 15, pp. 27872805, 2010.

[5] J. P. Conti, The internet of things, Communications Engineer, vol. 4,no. 6, pp. 2025, 2006.

[6] W. Shi and M. Liu, Tactics of handling data in internet of things,in Cloud Computing and Intelligence Systems (CCIS), 2011 IEEEInternational Conference on. IEEE, 2011, pp. 515517.

Fig. 10. Volcanic tremor event determined after digital processing signal

Fig. 11. High frequency volcanic tremor event determined after digitalprocessing signal

[7] G. Werner-Allen, J. Johnson, M. Ruiz, J. Lees, and M. Welsh, Mon-itoring volcanic eruptions with a wireless sensor network, in WirelessSensor Networks, 2005. Proceeedings of the Second European Workshopon. IEEE, 2005, pp. 108120.

[8] J. de Pedro Carracedo, Las tic en la prevencion de desastres naturales,in II Congreso de Computacion para el Desarrollo (COMPDES09).Acceso en: http://www. redusoi. org/docs/publicaciones/P11-Las 20TIC20en% 20la% 20prevencion% 20de% 20desastres% 20nat urales. pdf,2009.

[9] J. Zheng and A. Jamalipour, Wireless sensor networks: a networkingperspective. Wiley-IEEE Press, 2009.

[10] M. Kohvakka, M. Kuorilehto, M. Hannikainen, and T. D. Hamalainen,Performance analysis of ieee 802.15. 4 and zigbee for large-scalewireless sensor network applications, pp. 4857, 2006.

[11] I. Hammoodi, B. Stewart, A. Kocian, and S. McMeekin, A compre-hensive performance study of opnet modeler for zigbee wireless sensornetworks, pp. 357362, 2009.

[12] Y. Li, C. S. Chen, Y.-Q. Song, Z. Wang et al., Real-time qos supportin wireless sensor networks: a survey, in 7th IFAC International Con-ference on Fieldbuses & Networks in Industrial & Embedded Systems-FeT2007, 2007.

[13] K. S. Prabh, Real-time wireless sensor networks, Ph.D. dissertation,University OF Virginia, 2007.

[14] J. Gao, J. Hu, and G. Min, Performance modelling of ieee 802.15. 4mac in lr-wpan with bursty on-off traffic, in Computer and InformationTechnology, 2009. CIT09. Ninth IEEE International Conference on,vol. 2. IEEE, 2009, pp. 5862.

[15] J. Gao, J. Hu, G. Min, and L. Xu, Qos performance analysis of ieee802.15. 4 mac in lr-wpan with bursty error channels, in Mobile Ad-hocand Sensor Networks, 2009. MSN09. 5th International Conference on.IEEE, 2009, pp. 252256.

[16] M. Imran, A. Said, and H. Hasbullah, A survey of simulators, emulators

7

and testbeds for wireless sensor networks, in Information Technology(ITSim), 2010 International Symposium in, vol. 2, June, pp. 897902.

[17] J. Zheng and M. J. Lee, A comprehensive performance study of ieee802.15. 4, Sensor network operations, pp. 218237, 2004.

[18] Y. Xue, H. S. Lee, M. Yang, P. Kumarawadu, H. H. Ghenniwa, andW. Shen, Performance evaluation of ns-2 simulator for wireless sensornetworks, in Electrical and Computer Engineering, 2007. CCECE2007. Canadian Conference on. IEEE, 2007, pp. 13721375.

[19] F. Chen, N. Wang, R. German, and F. Dressler, Performance evaluationof ieee 802.15. 4 lr-wpan for industrial applications, in Wireless onDemand Network Systems and Services, 2008. WONS 2008. Fifth AnnualConference on. IEEE, 2008, pp. 8996.

[20] G. Tamilselvan and A. Shanmugam, Qualnet simulation of channel col-lision between ieee 802.15. 4 and ieee 802.11 b for various topologies,International Journal of Engineering and Technology, vol. 1, no. 2, pp.17938236, 2009.

[21] D. Borade and S. Laeeq, Performance and evaluation of ieee 802.15.4under different topologies with ad-hoc on demand distance vectorprotocol, in Electrical, Electronics and Computer Science (SCEECS),2012 IEEE Students Conference on, March, pp. 14.

[22] J.-S. Lee, Performance evaluation of ieee 802.15.4 for low-rate wirelesspersonal area networks, Consumer Electronics, IEEE Transactions on,vol. 52, no. 3, pp. 742749, 2006.

[23] J.-S. Lee, Y.-M. Wang, and C.-C. Shen, Performance evaluation ofzigbee-based sensor networks using empirical measurements, in CyberTechnology in Automation, Control, and Intelligent Systems (CYBER),2012 IEEE International Conference on. IEEE, 2012, pp. 5863.

[24] M. Petrova, J. Riihijarvi, P. Mahonen, and S. Labella, Performancestudy of ieee 802.15. 4 using measurements and simulations, in WirelessCommunications and Networking Conference, 2006. WCNC 2006. IEEE,vol. 1. IEEE, 2006, pp. 487492.

[25] T. R. Burchfield, S. Venkatesan, and D. Weiner, Maximizing throughputin zigbee wireless networks through analysis, simulations and imple-mentations, in Proceedings of the International Workshop on LocalizedAlgorithms and Protocols for Wireless Sensor Networks Santa Fe, NewMexico. Citeseer, 2007, pp. 1529.

[26] M. Kohvakka, M. Kuorilehto, M. Hannikainen, and T. D. Hamalainen,Performance analysis of ieee 802.15. 4 and zigbee for large-scalewireless sensor network applications, in Proceedings of the 3rd ACMinternational workshop on Performance evaluation of wireless ad hoc,sensor and ubiquitous networks. ACM, 2006, pp. 4857.

[27] G. Jimenez and M. Elena, Estudio de la eficiencia de encaminamientodel protocolo aodv en redes ad hoc inalambricas de gran escala, 2009.

[28] E. W. Weisstein, Schlafli symbol, from MathWorldA Wolfram WebResource. http://mathworld.wolfram.com/SchlaefliSymbol.html, visitedmarch 2013.

[29] M. Stehlk, Comparison of simulators for wireless sensor networks,2011.

[30] K. N. J. Malek, Trace graph-data presentation system for networksimulator ns, in Tools and Applications (ISAT 2003), sep. 2003.

[31] G. Werner-Allen, J. Johnson, M. Ruiz, J. Lees, and M. Welsh, Moni-toring volcanic eruptions with a wireless sensor network, pp. 108120,Jan 2005.

[32] G. Werner-Allen, K. Lorincz, M. Ruiz, O. Marcillo, J. Johnson, J. Lees,and M. Welsh, Deploying a wireless sensor network on an activevolcano, Internet Computing, IEEE, vol. 10, no. 2, pp. 1825, 2006.

[33] G. Werner-Allen, K. Lorincz, J. Johnson, J. Lees, and M. Welsh,Fidelity and yield in a volcano monitoring sensor network, pp. 381396, 2006.

[34] I. F. Londono Mancheno, Implementacion de un sistema de monitor-izacion de senales ssmicas del volcan cotopaxi empleando una red desensores inalambricos, 2011.

[35] R. K. Patra, S. Nedevschi, S. Surana, A. Sheth, L. Subramanian, andE. A. Brewer, Wildnet: Design and implementation of high performancewifi based long distance networks. vol. 1, no. 1, p. 1, 2007.