TNT: A Tactical Network Test Platform to Evaluate Military Systems Over Ever-Changing Scenarios

This paper addresses the challenge of testing military systems and applications over different communication scenarios with both network conditions and user data flows changing independently. We assume that systems developed to handle ever-changing communication scenarios are more likely to be reliable and robust during real military operations. Therefore, we propose the Tactical Network Test (TNT) platform to automate the evaluation of military systems and applications over real military radios using a reproducible test methodology. TNT has four main goals (i) the creation of QoS-constrained data flows; (ii) the execution of models to change network conditions; (iii) the automation of experiments to quantify the performance of military systems over ever-changing communication scenarios; and (iv) the monitoring of quantitative metrics and performing data analysis. Our platform was used to execute experiments in a VHF network by sending uniformly distributed data flows during seven different communication scenarios, either generated by a stochastic model or mobility models. The experimental results are used to discuss the military system’s performance by quantitative analysis using network metrics such as packet loss, delay, jitter, and data rate, and the test scenario characterization using mobility metrics such as speed, distance, and acceleration.

TNT platform, describing the models and the methodology to 96 create ever-changing communication scenarios in a test-bed 97 with real military radios. Section IV compares experimental 98 results using a set of patterns of link data rate change gen-99 erated with stochastic models and mobility models. Lastly, 100 Section V concludes the paper and lists future improvements. 101

102
This section discusses recent investigations introducing test 103 platforms for tactical networks together with the literature 104 using ever-changing communication scenarios as a method-105 ology to evaluate military applications. Therefore, Table 1 106 compares the main aspects to test and reproduce the proposed 107 methodologies, such as the methods used to create user data 108 flows and network conditions in the test platforms and appli-109 cations, the test environment deployed, and the reproducibil-110 ity of the proposed methodology in terms of documentation 111 and resources available.

113
The authors in [19] developed an emulation environment 114 for heterogeneous tactical networks together with operation 115 scenarios hosting several mobile units. The emulated test-116 bed, using Extendable Mobile Ad-hoc Network Emulator 117 (EMANE), can support different communication technolo-118 gies such as HF, UHF, VHF and SatCom due to its generic 119 Radio Frequency (RF) propagation model. Moreover, they 120 created different military scenarios called Anglova (Vignette 121 1 to 3) in the area of Fieldmont with 157 vehicles (network 122 node), in total, moving over the course of two hours. This 123 scenario was designed by military experts to be both realistic 124 and publicly available. 125 In [4], the authors proposed a Tactical Network Integration 126 Test Framework for simulation and emulation. Their solution 127 is composed of three test environments, namely simulation, 128 high-fidelity emulation, and scalable emulation. They started 129 with a high-fidelity test-bed as a baseline for a comparative 130 study. The study uses the same scenario for all three envi-131 ronments and tests them by comparing the performance and 132 ensuring consistency in the experimentation. The authors also 133 increased the network size to evaluate all test environments 134 again comparing each environment against each other.

135
In [5], the authors proposed an emulated test-bed to host 136 experiments with Software-defined Networks (SDN) solu-137 tions deployed in tactical scenarios [6], [20]. The emula-138 tion, also using EMANE, can execute various applications or 139 traffic generators in diverse SDN configurations. The SDN 140 layer can be configured to instantiate a specific network 141 architecture hosting SDN-capable nodes. Additionally, the 142 network scenarios change accordingly to the Anglova sce-143 nario [19]. The investigation reported in [21] developed a 144 test-bed for Software-defined Tactical Network (SDTN) tests 145 in an emulated environment using the Mininet-Wifi emula-146 tor. The experimental setup uses a specific SDN architec-147 ture composed of controllers in a distributed configuration, 148 hosting one global controller and multiple local controllers 149 changes mitigating radio buffer overflow and packet loss. 205 The proposed mechanism was evaluated over ever-changing 206 communication scenarios using different patterns of data rate 207 changes supported by real VHF radios [31]. 208 In summary, recent literature has a limited discussion 209 about testing military systems over ever-changing communi-210 cation scenarios in emulated and real environments. Meaning 211 that most studies use static or periodic changes and may 212 not develop reproducible and well-defined methodologies, 213 restricting/lacking the evaluation and characterization of test 214 scenarios, as listed earlier in Table 1. Different from the 215 literature, we designed a general-purpose platform, TNT, 216 designed to decouple the test of military systems into three 217 layers the application; the military systems; and the network 218 layer. Therefore, through defining models we argue that the 219 TNT provides a reproducible methodology allowing any vari-220 ation over the test scenario. Thus, the network can change 221 using stochastic/mobility models or real traces, and the user 222 data flow can be defined by either specific applications or 223 stochastic models. In addition, TNT also provided a prototype 224 to create link disconnection in real stationary radios with 225 wired antennas.

243
In this section, TNTs platform is described aiming to auto-

264
(1) Monitoring and data analysis: TNT collects the perti-265 nent logs (sender and receiver IP packets, radio buffer, 266 and radio modulation) during the experiment and pro-267 cesses them at the end of the experiment, triggered 268 by (0), for quantitative analysis of the experimental 269 conditions and the system performance.

270
(2) Military systems: in this phase TNT deploys a middle-271 ware, broker, or proxy at the nodes in a test environ-272 ment. This is the system subject of the performance 273 tests executed by TNT;

274
In the application layer (3), the user data flows are gen-275 erated by a model that creates different patterns of data 276 flows. From the network perspective (4), TNT creates a 277 topology that uses patterns of a data rate change or mobility 278 traces, simulating a wide range of network scenarios in a 279 communication area inspired by the wave propagation of 280 omnidirectional antennas. These scenarios are deployed in 281 real test-bed or emulated environments in order to test a given 282 military system (2). In this investigation, TNT is instantiated 283 to deploy network scenarios in a test-bed with real military 284 radios. The next sections discuss the main building blocks of 285 the TNT platform in detail. Moreover, the models to create 286 ever-changing network scenarios to test military systems, 287 as well as the shaping mechanism and the disconnection 288 prototype are available in a public repository. 1 The first step is to configure the experiment by using the two 291 models M A defining the user data flow, and M B defining 292 the network conditions and the test environment. The test 293 environment can be either created by an emulation software 294 or a real military test-bed; the experiments reported here were 295 executed in a real test-bed, as described later in Section IV. 296 At the application layer, TNT defines a model to create 297 different patterns of data flows, allowing the users to input 298 parameters such as the number of messages, size, and inter-299 message delays. Moreover, TNT uses model M B to configure 300 the changes in the network by defining how the radio link 301 data rate will vary during the experiments. This is done by 302 selecting and providing one of the three inputs (i) a sequence 303 of link data rates; (ii) a probability distribution (e.g. a Markov 304 chain) for the corresponding network states; or (iii) a mobility 305 trace (x, y, time) created by mobility models. Finally, TNT 306 defines the network topology by setting the link data rates 307 connecting them. The latter is set by parameters defining the 308 communication technology, which has a range (kilometers), 309 set of modulations describing nominal link data rates and the 310 overall end-to-end delay.   as follows: = (σ 0 , . . . , σ N −1 ) represents a sequence of 355 data rate changes over the set of system states S = {s 0 = 0, 356 s 1 = 1, s 2 = 2, s 3 = 3, s 4 = 4, s 5 = 5} representing the 357 nominal data rates {0.6, 1.2, 2.4, 4.8, 9.6} kbps supported by 358 the VHF radios (PR4G by Thales) in our laboratory and also 359 shown in Figure 2. Notice that the system state space S can 360 be defined in any desired way to match with the modulation 361 of the radios used in the test environment.

362
Moreover, the sequence of states can either be sampled 363 using probability distributions, such as used in [28], [29], and 364 [33], or manually defined by the user. The input parameter λ 365 defines the time interval for updating one system state s t n at 366 time t n to the next system state s t n+1 with n ∈ {0, . . . , N − 1}, 367 meaning that the model changes or disconnects the network 368 link in the test environment according to the time distribution 369 defined by λ. For example, assume that the movement starts 370 at t 0 = 0 in system state s t 0 = 2, the next state s t 1 = 0 and 371 λ = 10. Then, the link data rate between sender and receiver 372 will be set to 1.2 kbps starting at time 0 and remains the same 373 for exactly 10 seconds until the system changes to state 0, 374 meaning that the link will be disconnected for the upcoming 375 10 seconds. Next, the parameter C represents the commu-376 nication area with respect to the link quality defined by the 377 current data rate σ n ∈ . Moreover, different distributions 378 can be applied to sample the node positions over space using 379 the probability density function f .
containing the respective link data rate σ n . In other words, 393 M B 2 focuses on the transformation function, mapping the 394 p 0 , . . . , p N −1 ∈ to the ring-shaped areas C with respect to 395 a reference node n ref (x, y) ∈ R×R, describing a specific link 396 data rate. It is important to notice that TNT assumes that the 397 radios have an interface (e.g. Simple Network Management 398 Protocol (SNMP)) to change the link data rates (radio modu-399 lation) or add an attenuation (e.g. using a channel emulator) 400 to force the radio to change its modulation. {(s t 0 , p 0 , σ 0 ), . . . , (s t N −1 , p N −1 , σ N −1 )}, that describes a 406 given mobility pattern. TNT executes the following steps 407 to transform a sequence of data rates into a mobility trace: 408 First, we define a mapping of the system state S to a set 409 VOLUME 10, 2022  Given a node with position c(x, y), TNT implements 457 inverse transformation sampling to create the points 458 p 0 (x 0 , y 0 ), . . . , p N −1 (x N −1 , y N −1 ) of the raw mobility trace 459 R , sampled from the circular areas A i generated in the last 460 step. For this purpose, let us assume that we have a reference 461 node, which can be a base station or a neighboring node 462 sharing its location, with position n ref (x, y). Depending on 463 the link quality (system state) of the node with position 464 c(x, y) and the reference node n ref (x, y), TNT computes the 465 boundaries of the annulus A i , as shown in (1) Moreover, the inverse of F(X = d) in d is: Now, TNT uses inverse transform sampling, Eq. (4), 487 to generate values of X which are distributed according to F. 488 This works as follows: (1) Generate a random number u 1 from 489 the standard uniform distribution in the interval [0, 1]; (2) 490 find the inverse of the desired CDF F −1 (d) and (3) Once TNT sampled the distance X = d, it computes the 493 x and y coordinates of the point p(x, y) as shown in Eq. (5) 494 by first sampling the corresponding angle α with respect to 495 a uniformly distributed random number u 2 and calculating x 496 and y using α afterwards. As a result, it generates a sequence 497 of positions that we call Raw trace R . Moreover, TNT can 498 be specified to vary this model by restricting position p(x, y) 499 at time t n to be the nearest neighbor of the node position at 500 time t n−1 with distance d from center c(x, y), meaning that 501 TNT chooses α such that it minimizes the euclidean distance 502 of both positions. This approach allows the creation of new 503 traces, named Shortest trace, restricting the node movements 504 to the shortest distance using the same uniform coordinates 505 distribution.
Finally, in the last step, the model outputs a trace file 510 with node positions and link data rate σ n (node state) at a 536 Model M B 2 ( , C, n ref ) transforms mobility patterns into net-557 work states and is used to change the radio modulation or 558 cause link disconnections during the experiment. It should 559 be noted that TNT is designed to work with any tool that 560 implements mobility models and exports a trace file, such 561 as MobiSim [36] and BonnMotion [37]. To create network changes based on the mobility traces, 579 TNT uses the radio's SNMP interface to change the nom-580 inal link data rates and the link disconnection prototype to 581 create disruptions. Since our test-bed is using radios with 582 wired antennas inside a laboratory, it is not possible to cre-583 ate disconnections moving nodes away from the communi-584 cation range like in emulation platforms. Therefore, TNT 585 proposes a low-cost and easy-to-use link disconnection (state 586 s 0 = 0) prototype designed with a controller (Raspberry-587 Pi) and a coaxial relay. Notice that, TNT could also connect 588 to a channel emulator creating link disconnections using a 589 management interface. The documentation describing how to 590 build this prototype and its codes are available in a public 591 repository. 2 More precisely, the changes on the network are 592 done by two different interfaces depending on the node state 593 (i) connected: SNMP interface to the tactical radio (to change 594 its modulation) and (ii) disconnected: interface to the link 595 disconnection prototype 'cutting' the wired antenna using a 596 relay.

598
After introducing TNT's mechanisms to create ever-changing 599 user data flows and network conditions, the next step is to 600 design a military system in order to test the platform over 601 different scenarios as illustrated in Figure 1 (2). In previous 602 investigations, we observed that the limited buffer size of 603 military radios combined with the ever-changing network 604 scenarios and user data flows (say message size and time 605 distribution) results in buffer overflow [16], [38]. Therefore, we created a mechanism that can handle the radio buffer The proposed mechanism is explained below by instan-646 tiating three possible cases defined as follows: 1) suppose 647 the current data rate is d = 4.8 kbps, the max data rate 648 supported is max( d = 9.6 kbps), the min( d = 100 bps), 649 the buffer threshold is b = 50 %, the warning area is w b = 650 10 % and the current buffer occupancy is B = 20 %. 651 Then, the second condition is satisfied (|20 -50| > 10) 652 and the HTB bitrate will be set to the maximum data rate 653 supported by the radio. The goal is to fill up the buffer as fast 654 as the radio supports before the network condition gets worse; 655 2) now, the buffer occupancy reached B = 41 %, then the 656 third condition is satisfied (|41 -50| < 10) and the dequeue 657 rate is set to DQ r = 4.8 kbps, reducing the delivery rate from 658 the queue to the radio buffer; 3) finally, if B ≥ 50 %, then 659 the first condition will reduce the current DQ r in w b = 10% 660 until the buffer occupancy keeps lower than the threshold. 661 In short, the control mechanism changes the dequeue rate like 662 DQ r = { 4.8 kbps, 4.32 kbps, 3.88 kbps, 3.49 kbps, . . . , 663 100 bps} until the buffer usage B is close to the pre-defined 664 threshold. If the link data rate changes, the system will adapt 665 its dequeue rate DQ r in a similar way.

666
TNT was designed to test such a control mechanism 667 within military systems over a wide range of communication 668 scenarios.

670
Having everything needed to execute an experiment, the next 671 step is to deploy a monitoring and data analysis mechanism as 672 shown in Figure 1 (1). TNT process the data gathered from 673 both sender and receiver nodes and the trace files, analyz-674 ing and visualizing these data together. The monitoring and 675 data analysis phase was designed to allow for quantitatively 676 examining the experiment, helping to highlight issues regard-677 ing the test environment and quantifying the robustness of 678 systems such as routing protocols and tactical middlewares. 679 Including the monitoring of the end-to-end enforcement of 680 QoS requirements from command and control applications. 681 Next, we describe each step of this phase. This process was partly developed in our previous investi-684 gation [29] to collect all contextual data of both sender and 685 receiver nodes such as IP packets, and radio features (buffer 686 and modulation) and store them in a centralized database 687 which TNT uses to conduct further analysis. To quantify the military system performance, TNT process 690 the IP packets from both sender and receiver nodes. The 691 IP packets are sorted and combined as a function of time. 692 In sequence, TNT process the time series extracting network 693 metrics, such as data rate, radio buffer, delay, jitter, and 694 packet loss. Finally, TNT exports the statistics in two different 695 files: one compiling the overview of the experiment with 696 basic statistics, such as min, max, average, and standard 697   (here the adaptive shaping mechanism for user data flows) 736 on the sender node in order to avoid radio buffer overflow; 737 and (4) the monitoring and data analysis scripts which collect, 738 prepare and plot the experiment outputs.

739
The tactical network is composed of two VHF radios 740 (PR4G), with 128kb of buffer size and supporting five data 741 rates {0.6, 1.2, 2.4, 4.8, 9.6} kbps, each connected to a node 742 (sender and receiver, respectively). The radio antennas are 743 wired and connected to a link disconnection prototype in 744 order to create link disconnections, state {0}. Then, the net-745 work condition is changed using mobility traces as described 746 in Section III-C. Moreover, a server in a management network 747 is used with methods to conduct the data monitoring and 748 analysis, acquiring, processing, and plotting all logs. In the 749 next sections, we describe the user data flow and the patterns 750 of data rate changes used in the experiments, then discuss 751 the experimental results. The experimental setup with all 752 components used by TNT is listed in Table 2.  to ensure that the data will be sent during the whole mobility 785 pattern, which takes about 170 min. Besides, the goal is 786 to demonstrate TNT's functionalities, testing the queuing 787 discipline (part of a multi-layer military system) over a load of 788 packets that surely demands a store-and-forward mechanism 789 to successfully arrive at the receiver.   In order to better understand these mobility patterns, 827 Figure 6 shows the pattern of changes in a time series and 828 the overall statistics of each trace. Taking a look at the 829 changes over time in Figure 6a, we can see the difference 830 among the patterns, even though, they were generated with 831 the same probability matrix and sequence of states. This fact 832 shows that mobility should be considered in the discussion 833 of experimental results enriching the arguments to explain 834 such test scenarios, otherwise the experiments can be inter-835 preted or reproduced in different manners. In the MR trace, 836 we can observe rapid state changes, jumping from one data 837 rate to another. This pattern avoids any transitions between 838 states, making it hard for systems to predict the network 839 condition by monitoring the changes over time and acting to 840 reduce packet loss. The MF trace is different showing smooth 841 transitions between states by interpolating them. Based on 842 smooth movements, tactical systems can sense small and slow 843 network changes, that can be used as input to models designed 844 to predict movements that may increase the probability of 845 packet loss. Finally, the MSF trace combines smooth move-846 ments within short distances. Notice that this trace reduces 847 the node transitions because the node position is chosen not 848 considering all possible places inside the ring area but the 849 closest one from its last position.

850
The difference among these mobility patterns can be 851 observed from the trace statistics, as shown in Figure 6b. 852 This figure plots mobility statistics such as the distance 853    comprehension of the experiment's goals and limitations. For 879 example, these patterns can simulate the mobility of different 880 entities in a field, such as vehicles and humans patrolling a 881 given area. The MSF trace could represent the movement of 882 a vehicle with an average speed of 58 km/h in an uneven area, 883 requiring high variation on acceleration. The MR could char-884 acterize a non-human movement, e.g. an unmanned aerial 885 vehicle, covering larger distances and reaching high speeds. 886 Notice that these models can generate different test scenarios 887 by modifying a set of parameters that can be reproduced for 888 independent verification. 889

890
Before starting the experiment, TNT deploys the config-891 uration files for the traffic generator creating the IP data 892 flows, and the instance of model M B 1,2 creating changes in 893 the link data rates in our VHF network with real radios. 894 After the experiment ends, TNT plots six network metrics 895 in Figure 7. The metrics are the radio link data rate (Radio 896 DR), the data rate computed at the receiver side (DR), the 897 radio buffer (Buffer), the packet loss (P. Loss), the end-to-end 898 delay (Delay), and jitter (Jitter). Notice that the Radio DR 899 and Buffer metrics were acquired from the VHF radio through 900 SNMP and compiled together with the other metrics to show 901 the overall view of the experiment.

902
These experiments were designed to test the adaptive 903 shaping mechanism, as part of a multi-layer military system 904 described earlier in Section III-D, over different commu-905 nication scenarios. The Radio DR metric shows how the 906 network changes during a 170 min long experiment, which 907 ends when the node stops ''moving.'' However, the data 908 flow goes through the radio link until the packets' queue The shaping mechanism minimizes this issue by defining two   This section describes the experiments using model M B 2 , 970 which reuses well-known mobility models to create patterns 971 of link data rate change, as described earlier in Section III-C2. 972 TNT was set to use BonnMotion to generate four exemplary 973 mobility patterns, from four different models, namely GMs, 974 RWPs, MGs, and PRWs. Figure 8, shows how the node 975 movement is distributed over space and its respective network 976 states. Notice that model M B 2 is independent of the mobility 977 model generator BonnMotion and it supports any trace file 978 with a format node i = [x, y, time]. 979 Next, we briefly describe each model and the set of param-980 eters used to create a given mobility pattern. The goal is to 981 PRW [37] chooses new directions and speeds for the node 1005 to follow based on pre-defined probabilities. For example, 1006 Figure 8c shows the output of an instance of PRW using a 1007 Markov probability matrix to define probabilities distribution 1008 to move in any direction at a fixed speed or to remain in 1009 the same direction. This model also creates erratic movement 1010 with a given probability, simulating the unpredictable move-1011 ment of many entities in nature. Lastly, we used an instance 1012 of model RWP [44] to generate the mobility trace shown in 1013 Figure 8d. In this figure, the node starts without a change 1014 in direction for a certain period of time p t = 600 s, then 1015 it chooses at random the next destination. The node speed 1016 follows an uniform distribution and we defined the minimum 1017 speed around min s = 10 m/s ( 36 km/h) and the maximum 1018 speed of max s = 30 m/s ( 108 km/h). Once the node reaches 1019 the pre-defined checkpoints, such as cp 1 ( 5 km, 20 km), 1020 cp 2 ( 10 km, 22 km), and cp 3 ( 15 km, 24 km) it pauses for 1021 p t seconds before starting the process again. These patterns can represent different entities moving in a 1026 battlefield/scenario as described before, enriching the dis-1027 cussion and explanation of given system behavior based on 1028 a network scenario (node mobility). The difference among 1029 these patterns is also observed in the trace statistics shown in 1030 Figure 9b. Comparing the distance, speed, and acceleration 1031 features, it is noticed that the GM trace shows the most 1032 stable movements compared to the other patterns, with small 1033 variation around the mobility metrics with an average speed 1034 of about 47 km/h, acceleration about 0.005 km/h 2 averaged 1035 and distance traveled between positions about 0.27 km in 1036 average.

1037
The node at MG can reach 40 km/h and traveled dis-1038 tance about 0.12 km averaged with high averaged deceler-1039 ation about 0.9 km/h 2 , simulating a vehicle behavior in an 1040 urban area. The PRW trace shows long traveled distances 1041 about 0.44 km, compared with the last traces, at low speed 1042 about 31 km/h with no much variation in acceleration about 1043 0.0003 km/h 2 , all in average. For example, these attributes 1044 can be used to describe an unmanned patrol in a plateau 1045 terrain. Finally, the RWP trace introduces more variability 1046 in the traveled distances about 2.59 km averaged and reach-1047 ing maximum speed of 108 km/h. Moreover, this trace has 1048 specific checkpoints (areas) where it stays for a while before 1049 starts moving again, which can be characterized as vehicular 1050 movements, based on the speed and distances the entity can 1051 reach. Table 5 shows the numeric overview regarding the 1052 mobility traces MG, PRW, GM and RWP discussed in this 1053 section. 1054 2) TESTS WITH THE ADAPTIVE SHAPING MECHANISM 1055 Following the same methodology described in the experi-1056 mental results with model M B 1 . Here, the adaptive shaping 1057 mechanism is tested over the network conditions provided 1058 VOLUME 10, 2022   to GM with 101 s, RWP 112 s, and MG reaching 199 s. 1089 The Jitter metric of PRW reached the highest variation, about 1090 21 s, followed by MG with 12 s, GM with 7.4 s and RWP 1091 reaching about 5.7 s, all on average. Table 6 summarizes the 1092 quantitative results discussed in this section.

1094
This paper introduced and evaluated the TNT platform 1095 designed to test military systems over ever-changing com-1096 munication scenarios. TNT generates scenarios with different 1097 patterns of change for network conditions and user data 1098 flow, also collecting and analyzing data from experiments, 1099 and showing quantitative performance metrics. TNT wraps 1100 the necessary models to automate the evaluation of military 1101 systems and applications over real military radios. Therefore, 1102 TNT supports the creation of a variety of data flows by 1103 the definition of a general model (M A ) instantiated with a 1104 traffic generator tool, models (M B 1 and M B 2 ) for changing 1105 network conditions through stochastic models, transforming 1106 the sequence of network states (link data rates) in mobility 1107 patterns and mapping mobility traces to link data rates. As a 1108 result, mobility traces can be used to change the radio link 1109 data rates in stationary radios in a laboratory. TNT also 1110 includes a prototype to create link disconnections in radios 1111 with wired antennas.

1112
Addressing the problem of frequent changes in the link 1113 quality that may lead to a buffer overflow in military radios, 1114 TNT introduced an adaptive mechanism to shape the user 1115 data flow to the network conditions. Lastly, TNT's moni-1116 toring and data analysis scripts supported the discussion of 1117 experimental results over seven different scenarios, including 1118 plots and statistics, characterizing both the military system's 1119 performance using network metrics and the changes in the 1120 communication scenario using mobility metrics. 1121 We plan to extend TNT to support emulated network envi-1122 ronments with SDN-capable devices and military communi-1123 cation technologies, such as VHF, UHF, and SatCom. The 1124 goal is to increase the scale of the test environment and to 1125 improve military systems before deploying them in a close to 1126 a real network environment. Moreover, we plan to enhance 1127 the network-changing model M B 1 by introducing physical 1128 interference that simulates barriers along the mobility traces 1129 such as speed, distance, and non-linear movements. He is currently a Scientist at Thales Deutschland, 1320 Ditzingen, Germany. He is with the Secure Com-1321 munications & Information Systems (SIX), he has 1322 been attacking problems in computer networks and 1323 distributed systems with a particular interest in the 1324 performance bounds of tactical systems over ever-changing communication 1325 scenarios. He also visited universities at UTwente, The Netherlands, and 1326 NTNU, Norway. He worked as a Scientist at Fraunhofer FKIE, Germany. 1327 He has been rebuilding his own education following curiosity freely by 1328 reading books on physics, mathematics, and philosophy.