Comparison of Wireless Network Simulators with Multihop Wireless Network Testbed in Corridor Environment

Transcription

1 Comparison of Wireless Network Simulators with Multihop Wireless Network Testbe in Corrior Environment Rabiullah Khattak, Anna Chaltseva, Laurynas Riliskis, Ulf Boin, an Evgeny Osipov Department of Computer Science Electrical an Space Engineering Luleå University of Technology, Luleå, Sween, Abstract. This paper presents a comparative stuy between results of a single channel multihop wireless network testbe an the network simulators ns-2 an ns-3. We explore how well these simulators reflect reality with their stanar empirical raio moeling capabilities. The environment stuie is a corrior causing wave-guiing propagation phenomena of raio waves, which challenges the raio moels use in the simulators. We fin that simulations are roughly matching with testbe results for single flows, but clearly eviate from testbe results for concurrent flows. The mismatch between simulations an testbe results is ue to imperfect wireless propagation channel moeling. This paper reveals the importance of valiating simulation results when stuying single channel multihop wireless network performance. It further emphasizes the nee for valiation when using empirical raio moeling for more complex environments such as corriors. 1 Introuction Nowaays most of the research in the fiel of wireless networking is base on network simulators. Simulators are attractive for researching network protocols an mechanisms since they allow creating controlle an reproucible environments. Creating such an environments in real test bes is both expensive an time consuming. Real prouction networks at the same time often o not allow to obtain repeatable ata sets neee for research analysis. Various network settings an large parameter ranges can be teste through simulations at low effort since creation an moification of network scenarios as well as ata gathering are easy. In this paper we stuy how well simulations reflect the reality with commonly use empirical moels of wireless propagation channel that requires little configuration an are memory an computationally sparse. In particular, we explore isparities between simulations in ns2 1 an ns3 2 an testbe results for a multihop wireless network locate in a corrior. This environment challenges the wireless propagation channel moel present in the network simulators. 1 ns-2 NetworkSimulator. Online. Available: 2 ns-3 Network Simulator. Online. Available:

2 The performance of network protocols in a testbe is affecte by wireless channel properties that epen on the physical environment, location an mobility of the noes, an the external interference. Accurate wireless channel moeling for simulations is known to be ifficult. The commonly use wireless propagation channel for path loss in simulators is empirically moele, an path loss is compute epening on istance between transmitter an receiver. Consequently, accumulative interference cause by hien terminals concurrent transmissions, an spatial reuse ratios of testbe network may not be correctly represente by the simulators. Due to these reasons simulation results often o not match perfectly with the testbe results [2, 5]. The moeling of wave-guiing propagation phenomena of raio waves in corriors as well as moeling of the losses cause by reflections, iffraction an scattering of raio waves are more accurately capture by eterministic channel moeling methos [7]. The eterministic wireless channel moeling, e.g base on ray tracing techniques require the exact knowlege of location, shape, ielectric an conuctive properties of all objects in the environments an it also requires extensive computational efforts for accuracy. Thus such moels are site specific. In aition these moels also consierably increase simulations run time [12]. The aim of our paper is to make the wireless network researchers aware of the ifferences of the simulations from real wireless testbe. The ifferences are mainly cause by empirically moele wireless propagation channel of the simulators. Empirical wireless propagation channel moels of the simulators are simple from implementation perspective but they o not cover all the properties of the wireless propagation channel such as losses ue to reflections, iffraction, scattering an penetration of the raio waves. We emonstrate the ifferences between simulations an testbe for two specific scenarios compose of single an concurrent flows transmissions over a single multihop path. We fin that for single flow transmissions over multiple raio links, ns2 an ns3 simulations roughly match the testbe results. Deviations between simulations an the testbe results are explaine by the wireless propagation channel moels use in the simulations are not accurately reflecting the accumulative interference cause by hien terminals concurrent transmissions an the spatial reuse ratio 3 of the testbe network. This shortcoming of the wireless propagation channel moel becomes more evient for simultaneous flows transmissions, which have higher strength of accumulative interference than single flow. Simulations inicate consierably worse matching of the throughput fairness of simultaneous flows with testbe results, which reveals that the wireless propagation channel moels of the simulators are not correctly representing the wireless channel properties in the corriors. The article is organize as follows. Section II presents a brief overview of the backgroun an relate works. Section III presents the etails an specifications of the testbe network an experiments. Section IV gives the etails of the simulation setup an also provies an overview of path loss an multipath faing moels. Section V presents the comparison of the experimental results with simulations. Section VI conclues this comparative stuy. 3 Spatial reuse ratio is the total number of concurrent transmissions accommoate in network.

3 2 BACKGROUND AND RELATED WORK Network simulators ns-2 an ns-3 are e-facto stanar simulation tools in the acaemic networking research community. Simulations in ns-2 are constructe with C++ coe an OTcl scripts; the former provies moeling of applications, simulation noes, communication channels an other mechanisms involve in networking, while the latter is use to control simulations an efine aitional features, for example the network topology. Simulations in ns-3 are fully base on C++, but can also be create with Python. The ns-3 simulator was evelope from scratch an cannot irectly use the coe evelope for ns2. Many objects are porte from ns-2 to ns-3 but not all, an hence ns-2 incorporate capabilities not present in ns-3. However, ns-3 has capabilities not implemente in ns-2 such as support for multiple interfaces on noes, use of IP aressing an closer resemblance with the TCP/IP moel, an more etaile a/b/s moels. The accuracy of wireless channel moels for simulations naturally etermines the quality of the outcome. It coul be expecte that the more etaile moeling of the IEEE MAC protocol in ns-3 woul result in more accurate results for certain network scenarios. Obviously it is expecte that the simulations results may eviate from the reality. It is therefore important to unerstan the egree of reality reflection by the simulators. Other stuies presenting comparisons between simulations of IEEE base networks an testbes inclue [9], which presents a comparative stuy between an IEEE a base testbe an three network simulators (ns-2, QualNet an OPNET). It aims to assess the relevance of these simulators in inoor an outoor environments. The simulation results match to some extent with the testbe. The authors highlight that tuning of physical layer parameters an selecte propagation moels have great impact on the results. This stuy is conucte for a single hop network an no comparison with ns-3 simulations is presente. In [1] the authors present a valiation stuy of the IEEE b MAC moel in ns- 3 by comparing simulations with testbe results. The stuy shows that ns-3 simulations nearly match with reality after proper tuning of the evices in the testbe. It is also shown that for mismatching between the simulation an testbe results, simulator is not always wrong but specific selection an configuration of the evices in the testbe can be culprit. However, in the testbe wireless channel propagation effects on measurements are ignore because the communication between the evices is via coaxial cables. The authors in [14] point out the isparities between a wireless network tesbe an ns-2 an Qualnet. The isparities are explore base on antenna iversity, path loss, multihop, transmission rate, interference an routing stability. However, ns-3 simulations have not been taken into account an intra-path interference 4 is iscusse only for a single flow traffic over a linear multihop network. 4 Interference between the packets of the same en-to-en flow ue to hien terminals.

4 Fig. 1. Testbe: Layout of noes in corriors. 3 Testbe an experiments escription We built an IEEE802.11b base multihop wireless testbe in an inoor corrior environment. The network consists of eight noes place as illustrate in Fig. 1. The logical topology is a chain, the placement of noes ensures the line-of-sight communications with the immeiate neighbors. All noes are Intel Pentium 4 base esktop PCs with 2.40 GHz processor, cache size 512 KB, RAM memory 256 MB an six USB 1.0 ports supporting the ata transfer rate of 12 Mb/s. For wireless connectivity each noe is equippe with D-Link DWL-G122 wireless USB aapter 5 with an omni-irectional antenna. The operating system is Linux (kernel ) an the WLAN river is p54 6. At the MAC layer we switche off the options for frame fragmentation, ynamic rate aaption an isable the RTS/CTS exchange. The transceivers operate on channel 3, the transmit power at each noe was set to 18Bm, the physical channel ata rate is set to 11Mb/s. We experimente with TCP traffic generate by Iperf (version 2.0.8) traffic generator. We use TCP-Cubic configure with the efault settings. The routes were configure statically in orer to eliminate the effect of routing protocols on network performance [8]. In the testbe we conucte two types of experiments. The first experiment (further on referre to as experiment-1) was performe with a single flow running over a ifferent number of hops. The secon experiment (further on referre to as experiment-2) was conucte with multiple TCP flows running concurrently over ifferent number of 5 DWL-G122 High Spee 2.4GHz (802.11g) Wireless USB Aapter. Online. Available: http: // 6 Online.Available:

5 (a) Layout of Experiment-1. (b) Layout of Experiment-2. Fig. 2. Setup of experiments in the testbe. hops. To obtain the results from both experiments, we use tcpump (version 3.9.8) to capture all the traffic generate in the network an measure the per-flow throughput. The setup for experiment-1 is shown in Fig. 2(a). The experiment consists of seven scenarios with ifferent number of wireless hops for the monitore flow. In orer to capture the effects of multipath faing an the accumulative interference on the network performance each scenario was repeate six times by changing the position of the source an estination noe along the multihop chain for a given number of hops. The uration of each trial is three minutes. In experiment-2 (see Fig. 2(b) ) the layout of the noes an the network specification ha been kept the same as in experiment-1. The experiment consists of six scenarios. In all scenarios the traffic is alway generate from noe 8. We start experimenting with two flows of one an two hops running in parallel. In each subsequent scenario we a flows as shown in the figure. As a result in the sixth scenario we experiment with seven concurrent TCP flows. We perform two trials for each scenario an recor an average value of the per-flow throughput. The uration of each experiment is two minutes. The throughput values are use to compute a fairness inex as explaine below. 4 Simulation setup We replicate the testbe experiments in ns-2 an ns-3 simulators. Firstly, we configure the parameters of the simulators with the corresponing values in the testbe. In particular, the transmission power, characteristics of the antenna an the corresponing transmission ranges are set accoring to the specification of the D-Link DWL-G122 wireless USB aapter. Accoring to the evice s ata sheet the transmission range is set to 100 m. Accoring to [11] the carrier sensing range in commercial wireless cars is twice or more than the transmission range. On the transport layer we use TCP-Cubic as in the testbe experiments. Note that while the implementation of TCP-Cubic in ns-2 is a simulator-specific, ns3- links the real implementation from Linux via the Network Simulator Crale (NCS). Special attention was pai to the proper configuration of the path loss an multipath faing moels in orer to reflect the raio environment of the testbe. In the inoor environment, the propagation of raio waves is mainly affecte by two types of losses: the

6 path loss an the loss ue to small an large scale faing. The small scale faing arises ue to the multipath propagation effect an the large scale faing is ue to the shaowing effect. The moel which closely reflects the path loss in the inoor environment is the log-istance path loss moel [10]: L p = L n log X θ. (1) In (1) n is the path loss istance exponent, 0 is the reference istance (1 m), is the istance in meters between the transmitting an the receiving noes, L 0 is the reference path loss at the reference istance (B), L p is the path loss (B) an X θ is a log-normally istribute ranom variable (B) with stanar eviation σ an zero mean escribing the attenuation cause by the obstacles ue to shaowing effect. Note that the value of n epens on the operating frequency an the characteristics of the propagation environment. In the case of the inoor environment, the type of the construction material an the position of the noes within the builing. In corriors ue to wave-guiing propagation phenomena of raio waves, n takes values in the range 4π [1.3, 1.9] at 2.4 GHz [3, 6]. From Friis propagation loss moel L 0 = 20 log 0 10 λ where λ is the wave length in meters. Note that in our testbe there is no large scale faing ue to line-of-sight communication between the ajacent noes, therefore, X θ = 0 an (1) reuces to L p = L n log (2) However, the small scale faing exists in the corriors ue to multipath propagation. In corriors the small scale faing is escribe by Nakagami istribution [13, 15]. The probability ensity function for the Nakagami m-istribution is pf r = 2 ( m ) m ( r 2m 1 exp m Γ (m) ω ω r2). (3) In (3) r 0 is the amplitue of the receive signal, Γ (m) is the Euler s Gamma function, ω = r 2 is the mean square receive power an m = ω2 is the faing epth, (r 2 ω) 2 where m 1/2. For m = 1, the Nakagami m-istribution becomes Rayleigh istribution. In ns-3 the log-istance path loss an Nakagami faing moels are implemente separately. Both moels can be use an configure iniviually. In ns-2 the log-istance path loss moel is not yet implemente, however, the Nakagami faing moel is implemente together with three-log-istance path loss moel. The three-log-istance path loss moel of ns-2 is ifferent from the log-istance path loss moel with three istance fiels namely near, mile an far. Each fiel has ifferent path loss exponent. For the three-log-istance moel a fourth istance fiel is also efine from 0 to near istance, however, the loss over this fiel is zero. The limits of four istance fiels along with their corresponing path loss exponents are explaine as 0 } {{ } 0 }{{} 1 }{{} 2 }{{} 0 n 0 n 1 n 2

7 So each fiel starts at the en of the preceing one an hence the resultant three-logistance path loss moel is a continuous function of the istance: 0 0 L n 0 log L p = < 1 L n 0 log n 1 log < 2 L n 0 log n 1 log n 2 log In (4) n 0, n 1, n 2 are the path loss istance exponents an 0, 1, 2 are three istance fiels(meter). The Nakagami faing moel in ns-2 efines three parameters m for three istance fiels as 0 } {{ } 1 }{{} 2 }{{} m 0 m 1 m 2 The efaults values of istances an faing parameters m in ns-2 are 1 = 80 meter, 2 = 200 meter, m 0 = 1.5 an m 1 = m 2 = In the testbe the maximum istance between the ajacent noes per hop is less than 80 meter, so ns-2 is using log-istance path loss moel part of (4). In orer to fin a better match of the path loss an multipath faing with the real testbe the simulations are conucte for five combinations of the path loss exponent n = 1.9 an five Nakagami faing parameters m i.e. 1.5, 1.75, 2.0, 2.25 an Note that oing simulations with higher values of the faing parameter m is not realistic because the higher m means stronger LOS component of the propagation moel. This is not the case with the D-Link evices use in the testbe, which are equippe with omni-irectional antennas. The next section report the results of the analysis. (4) 5 Comparative analysis 5.1 Experiment-1: testbe vs. simulations Fig. 3(a) an Fig. 3(b) show ns-3 an ns-2 simulations of experiment-1 along with corresponing testbe results of experiment-1. As expecte the TCP throughput ecreases with increasing number of hops. This is because of the increase in accumulative interference ue to hien terminal problem an the ecrease in spatial reuse ratio of the network ue to the expose terminal problem. In case of testbe results, the most severe effect of accumulative interference an lower spatial reuse ratio on the throughput is observe in the five hops scenario. This severe effect is ue to the specific positions of the hien an expose terminals in the connecte corriors. It is, however, observe that the throughput of the six hops is higher than that of the five hops ue to its higher spatial reuse ratio an hence lower effect of the accumulative interference. Similarly the seven hops has higher throughput as compare to both six an five hops because of its further higher spatial reuse ratio. The higher spatial reuse ratios in the six an seven hops scenarios are attaine ue to the specific positions of the noes in the corriors. Fig. 3(a) shows that ns-3 simulate results are matching with the results from the testbe in the single hop scenario for faing parameter m = 2.0 while iverging in

8 (a) ns-3 results. (b) ns-2 results. Fig. 3. TCP throughput in experiment-1 versus ns-3 an ns-2 results for path loss exponent n = 1.9 an ifferent Nakagami parameters m. all other scenarios except for the six hops scenario. There the simulation results of all faing parameters m are almost ientical with the testbe results. Fig. 3(b) shows that for the single hop scenario the ns-2 simulation results are almost ientical with that of the testbe for faing parameter m = 2.0. The figure shows that ns-2 simulate results in three an four hops scenarios are matching with the testbe for various faing parameters m. Notably, simulations of ns-2 an ns-3 fail to reflect the higher spatial reuse ratio behavior of the six an seven hops than five hops like the testbe. The simulation results from both ns-2 an ns-3 show that none of the faing parameters m has a persistent match with the testbe results in all multihop scenarios. We also observe that in contrast to ns-3, the ns-2 simulations have a closer match with the testbe results except for six an seven hops cases, where ns-2 results are iverging from the testbe larger than ns-3. It is observe as well that the throughput of a single hop TCP flow is higher in the testbe than the simulate results for certain values of faing parameter m. This is ue to wave-guiing signal propagation phenomena not correctly capture by the propagation moels in the simulators. It is, therefore, har to conclue which of the two simulators closer reflect the reality except for stating that both simulators give a rough match of the testbe results. 5.2 Experiment-2: testbe vs. simulations In experiment-2, from the simulations of each simulator, we are getting five plots for the average throughputs (over all scenarios) for five Nakagami parameters. So for sake of simplicity of explanation an limitation of pages of the article, we present comparison of the results with a concise performance metric call throughput fairness inex. We fin that fairness inex behaviours of both the simulators are quite ifferent from the testbe, which implies that average throughputs behaviours of simultaneous flows of simulations are also ifferent from those of experiment-2 of the testbe. The etails of the throughput fairness inex are given as follows.

9 In experiment-2 we compare the performance of the simulators with the testbe using Jain fairness inex f s (5). The inex takes values between 0 an 1. In (5) X (s) i is the network s throughput share obtaine by i th flow an s is the number of simultaneous flows. f s = ( s ) 2 i=1 X(s) i s s i=1 (X(s) i ). (5) 2 In orer to use the inex in the case of flows with unequal characteristics we have to relate the actual measure throughput with a throughput share of the flow uner ieal sharing conition [4]. Therefore, X (s) i is compute as in (6). There a (s) i is the actual throughput of the i th flow measure in simulations an in the testbe an (s) k = T k s is the throughput share uner ieal sharing conitions over k hops. It is compute by iviing the throughput T k of a single flow over k hops measure in experiment-1 by s. X (s) i = { a (s) i (s) k if a (s) i 1 otherwise < (s) k As we observe from Fig. 4(a) an Fig. 4(b) the fairness inices obtaine from the ns-2 an ns-3 simulators have opposite trens for ifferent values of simultaneous flows an the number of hops. It is observe from ns-3 simulations in the Fig. 4(a) that the fairness inex is ecreasing in scenarios one to four. The lowest fairness inices are observe in scenarios five, six an seven for certain values of faing parameters m. Like in the testbe, ns3 simulations of scenarios five an greater show an increasing trens in the fairness inices for ifferent faing parameters. Clearly the fairness inex obtaine from ns3 simulations is lower than the one in the testbe for all values of the faing parameters. It is however worth pointing out that the overall behavior of the inex for Nakagami parameters m = 1.5 an m = 1.75 matches the inex s behavior in the testbe in scenarios three to seven. Looking at the results from ns-2 simulations in Fig. 4(b) we observe that the fairness inices increase with increasing number of simultaneous flows an hops. In scenario three the fairness inex in simulations matches exactly the values in the testbe. Although the absolute values of the inex obtaine from ns-2 o not significantly eviate from the measure in the testbe, however the overall evelopment of the inex is ifferent from testbe. It is to be note that wave-guiing propagation phenomena of raio waves in the testbe are present in both single an simultaneous flow scenarios. However, in the simultaneous flow scenarios the probability of concurrent transmissions along the multihop path is higher than single flow scenarios. This results in higher value of the accumulative interference in experiment-2. We know that wave-guiing propagation phenomena of raio waves reuce the signal strength loss as compare to common raio waves propagation in space. So ue to wave-guiing propagation, the accumulative interference has higher range to affect the esire reception of the signal along the multihop path. Hence the eviations of the simulations of simultaneous flow scenarios from testbe are larger than single flow scenarios. (6)

10 (a) ns-3 results. (b) ns-2 results. Fig. 4. Jain fairness inex in experiment-2 versus ns-3 an ns-2 results for path loss exponent n = 1.9 an ifferent Nakagami parameters m. Overall, as in the case with experiment-1 none of the simulators was able to exactly reprouce the performance of the testbe. Partially it epens on simulator specific implementation of protocols on MAC an (or) Transport layers. ns-2, for example, uses own implementation of TCP-Cubic congestion control. We however observe a better match of the network behavior prouce by ns-3 simulator which uses native Linux implementation of TCP. The major problem in our opinion comes however from inability of the simulator s propagation moels to capture all signal impairments mechanisms in the particular communication environment. The corrior environment of the tesbe exposes strong wave-guiing propagation phenomena, which places a great impact on the accumulative interference an the spatial reuse ratio. 6 Conclusion This paper presents a comparative stuy between testbe an simulations of the network simulators ns-2 an ns-3. The testbe is multihop wireless network eploye in corriors in a non-linear chain topology, which challenges the commonly use empirical moels of wireless propagation channel that are currently available in these simulators. The experiments one for this stuy inclue single an concurrent flows transmissions over a single multihop path. The goal is to explore how well these simulators reflect the reality represente by this testbe carrying those flows. Our simulations roughly match with testbe results for single flow transmissions, which cause only limite accumulate interference an allow for goo spatial reuse ratio of the network. Simulations eviate however more clearly from testbe results for simultaneous flows transmissions. These transmissions increase the accumulative interference compare to single flow transmissions an thereby ecreases the spatial reuse ratio of the network. In particular, for simultaneous flows transmissions simulations inicate consierably worse fairness between flows compare to testbe results. This reveals that the wireless propagation channel moels of the simulators are not correctly

11 representing the wireless channel properties in the corriors, especially in scenarios involving accumulate interference in ifficult environments such as corriors. Deterministic wireless channel moeling for example base on ray tracing techniques can better capture reality into simulators, but require exact knowlege of location, shape, ielectric an conuctive properties of all objects in the environments an it also requires extensive computational efforts for accuracy. This complexity motivates the use of empirical raio moeling although its shortcomings in correctly moeling complex environments. Our paper emphasizes the nee for valiation when using such moeling to stuy single channel multihop wireless network performance for more complex environments. Future work inclue to exten this stuy by exploring how results match between simulations an real multihop wireless networks with linear chain topology for other inoor environments as well as outoor environments. Thereby we can further characterize possible iscrepancies important to be aware of when using simulations to preict performance in real networks. References 1. Balo, N., Requena-Esteso, M., Nú nez-martínez, J., Portolès-Comeras, M., Nin-Guerrero, J., Dini, P., Mangues-Bafalluy, J.: Valiation of the ieee mac moel in the ns3 simulator using the extreme testbe. In: SIMUTools 10: Proceeings of the 3r International ICST Conference on Simulation Tools an Techniques. pp ICST (Institute for Computer Sciences, Social-Informatics an Telecommunications Engineering), ICST, Brussels, Belgium, Belgium (2010) 2. Cavin, D., Sasson, Y., Schiper, A.: On the accuracy of manet simulators. In: POMC 02: Proceeings of the secon ACM international workshop on Principles of mobile computing. pp ACM, New York, NY, USA (2002) 3. Giannopoulou, K., Katsareli, A., Dres, D., Vouyioukas, D., Constantinou, P.: Measurements for 2.4 GHz sprea spectrum system in moern office builings. p. 21 (2000) 4. Jain, R., Chiu, D., Hawe, W.: A quantitative measure of fairness an iscrimination for resource allocation in share systems. ec research report TR-301 (1984) 5. Kotz, D., Newport, C., Gray, R.S., Liu, J., Yuan, Y., Elliott, C.: Experimental evaluation of wireless simulation assumptions. In: MSWiM 04: Proceeings of the 7th ACM international symposium on Moeling, analysis an simulation of wireless an mobile systems. pp ACM, New York, NY, USA (2004) 6. Lu, D., Rutlege, D.: Investigation of inoor raio channels from 2.4 GHz to 24 GHz. vol. 2, pp vol.2 (jun 2003) 7. Molisch, A.: Wireless Communications. John Wiley & Sons (2005) 8. Osipov, E., Tschuin, C.: Evaluating the effect of a hoc routing on tcp performance in ieee base manets. In: Koucheryavy, Y., Harju, J., Iversen, V. (es.) New2AN 06: Proceeings of the 6th International Conference on Next Generation Teletraffic an Wire/Wireless Avance Networking. Lecture Notes in Computer Science, vol. 4003, pp Springer Berlin / Heielberg (2006) 9. Rachei, A., Lohier, S., Cherrier, S., Salhi, I.: Wireless network simulators relevance compare to a real testbe in outoor an inoor environments. In: IWCMC 10: Proceeings of the 6th International Wireless Communications an Mobile Computing Conference. pp ACM, New York, NY, USA (2010)

12 10. Rappaport, T.S.: Wireless Communications Principles an Practice. Prentice-Hall, Upper Sale River, NJ (2002) 11. Razak, S., Kolar, V., Abu-Ghazaleh, N.B., Harras, K.A.: How o wireless chains behave?: the impact of mac interactions. In: MSWiM 09: Proceeings of the 12th ACM international conference on Moeling, analysis an simulation of wireless an mobile systems. pp ACM, New York, NY, USA (2009) 12. Schmitz, A., Wenig, M.: The effect of the raio wave propagation moel in mobile a hoc networks. In: MSWiM 06: Proceeings of the 9th ACM international symposium on Moeling analysis an simulation of wireless an mobile systems. pp ACM Press, New York, NY, USA (2006) 13. Sheikh, A., Abi, M., Hanforth, M.: Inoor mobile raio channel at 946 mhz: Measurements an moeling. pp (may 1993) 14. Tan, K., Wu, D., (Jack) Chan, A., Mohapatra, P.: Comparing simulation tools an experimental testbes for wireless mesh networks. pp. 1 9 (jun 2010) 15. Tarng, J., Liu, W.S., Huang, Y.F., Huang, J.M.: A novel an efficient hybri moel of raio multipath-faing channels in inoor environments. Antennas an Propagation, IEEE Transactions on 51(3), (mar 2003)