Simulation and Analysis of Mobile Ad-hoc Network Technology in...

ABSTRACT

Amobile ad hoc network (MANET)

is an autonomous communicationssystem of mobile nodes equipped

with radio transmitters and receivers. Thisresearch explores three critical challengesfaced by communications planners in em-ploying MANET technology within the USMarine Corps infantry battalion. First, weexamine and quantify the ability of MANETsto support communications between highlymobile units operating in potentially ruggedterrain over long distances with relativelylow-power radios. We also analyze the abil-ity of MANETs to use intermediate nodesto overcome the inherent range limitationsof higher frequencies. Finally, we considerthe challenge of allocating bandwidth toMANET systems to enable sufficient through-put rates. To explore these challenges, weconduct a rigorous comparative analysisusing various network simulation and opti-mization techniques. We develop a networkformulation to model key aspects of com-munications systems, and then simulateand gauge network performance in environ-ments ranging from low-fidelity, theoreti-cal representations to realistic, high-fidelitycombat scenarios.

We quantify the benefit that MANETscan provide to tactical communications net-works over traditional point-to-point net-works. We also quantify the value of theuse of unmanned aircraft systems (UASs)as airborne nodes in MANETs, a capabilityespecially useful in communications sce-narios involving rugged terrain and largedistances. We also find that due to MANETfragility in high-loss environments, tacticsmay need to be modified to support thefull use of MANET communications. To ourknowledge, we are the first to rigorouslyexamine and quantify the value of MANETtechnology within the Marine Corps in-fantry battalion.

INTRODUCTION

Description of ProblemTactical forces within the US Marine

Corps (USMC) are becoming increasingly de-pendent on the rapid, reliable transfer of in-formation throughout the battlespace. MarineCorps Vision and Strategy 2025 (USMC, 2008)

states ‘‘the Marine Corps will integrate C2[command and control] and ISR [intelli-gence, surveillance, and reconnaissance] ca-pabilities down to the squad level,’’ and‘‘we will aggressively pursue integratedmicrotechnologies, such as a secure commu-nications personal data apparatus that com-municates via the spoken word, data, andimagery.’’ The Marine Corps document AConcept for Enhanced Company Operations(2008) emphasizes ‘‘support to highly mobileforces with on-the-move/over-the-horizoncommunications for disparate tactical nodes,’’and states to achieve this, ‘‘tactical unitsmust gravitate from push-to-talk radio sys-tems to mobile ad-hoc mesh networking.’’

A mobile ad hoc network (MANET) isan autonomous communications system ofmobile nodes equipped with radio transmit-ters and receivers. These nodes may moveand connect in wireless, dynamic, multihoptopologies, and exhibit self-learning, self-healing behavior (Corson and Macker, 1999;Aggelou, 2005)—that is, individual nodesmay automatically connect and disconnectfrom a MANET without any user interac-tion. A MANET system comprises physicalradios and the associated networking proto-cols, waveforms, and modulation schemes.We assume all nodes within a MANET canserve as the source and/or destination ofcommunications traffic, and can functionas intermediate nodes to route communi-cations traffic from source to destination.MANET nodes may physically connect toclient devices (such as laptops), or serveas user terminals themselves. In this way,MANETs are similar to client mesh wire-less mesh networks (WMNs) (Zhang et al.,2006, pp. 564–567), where client devices per-form routing functions.

Existing tactical radio systems withinthe USMC infantry battalion include theSingle Channel Ground and Airborne Ra-dio System (SINCGARS) and various otherradios operating in the high frequency (HF),very high frequency (VHF), and ultra-highfrequency (UHF) ranges. In contrast toMANETs, these existing systems can func-tion only as point-to-point (PTP) networks:that is, nodes must communicate directlyand cannot function as intermediate nodes(Hong et al., 1999). Although some radiosystems allow nodes to function as retrans-mitters, in such a configuration nodes aresimply rebroadcasting received traffic andnot operating as part of a MANET.

Simulationand Analysisof MobileAd hocNetworkTechnologyin the USMarineCorpsInfantryBattalion

Paul NicholasJosh PepperCarol WeaverDavid GibbonsMark Muratore

Marine Corps CombatDevelopment Center,

[email protected],[email protected],[email protected],[email protected],[email protected]

APPLICATION AREAS:Battle Management/Command and Control;Land and ExpeditionaryWarfare; Modeling,Simulation, andWargaming

OR METHODOLOGIES:NonlinearProgramming; NetworkMethods; Simulation

Military Operations Research, V18 N4 2013, doi 10.5711/1082598318419 9Military Operations Research, V18 N4 2013, doi 10.5711/1082598318419 Page 19

The Marine Corps is currently developing,purchasing, and fielding new radio devices ca-pable of forming MANETs to connect widelydispersed nodes with data communications ca-pabilities. These systems include the PRC-117Gwideband multiband tactical radio, radios be-longing to the (now defunded) Joint TacticalRadio System (JTRS) family, the ExpeditionaryCommunications (EXCOMM) vehicle, and others(Harris Corporation, 2011; Marine Corps Sys-tems Command, 2011). Yet, the service has notextensively studied the tactical considerationsof the use of MANET architectures versus tradi-tional tactical architectures, nor the most effec-tive employment of these systems in a tacticalenvironment.

This research explores three critical chal-lenges faced by communications planners inemploying MANET technology within theUSMC infantry battalion. First, tactical forcesare increasingly operating in a geographicallyseparated manner. Such dispersed units requirecommunications networks able to dynamicallyovercome the challenges of wirelessly connect-ing highly mobile units operating in potentiallyrugged terrain over long distances with rela-tively low-power radios (Goulding, 2009). Weexamine and quantify the ability of MANETsto support operations in these demanding en-vironments.

Communications planners must also con-sider the challenges of signal attenuation.MANETs typically operate at higher radio fre-quencies (RFs) than traditional PTP networks.These higher frequencies enable greater through-put rates, but in general are subject to greater sig-nal attenuation than lower frequencies, and willthus propagate a shorter distance (all otherthings being equal) (Singal, 2010, pp. 37–39).However, unlike traditional PTP networks,MANETs may use intermediate nodes to relaytraffic. We analyze the ability of MANETs to useintermediate nodes to overcome the inherentrange limitations of higher frequencies.

Finally, we consider the challenge of allo-cating bandwidth to MANET systems to enablesufficient throughput rates. The throughput ca-pacity of a wireless link is a function of bothbandwidth and received signal strength (RSS)(Shannon, 1949). Bandwidth is the range of theRF spectrum allocated for use in Hertz (Hz),

and RSS is the RF power of the received signal,as measured at the receiver in dBm or watts.Since MANETsystems typically operate at higherfrequencies and are thus subject to greater sig-nal attenuation, greater bandwidth is requiredfor MANET systems to provide throughputrates equal to lower-frequency systems. We ex-amine and compare the bandwidth require-ments for lower-frequency PTP systems andhigher-frequency MANET systems.

To explore these challenges, we conduct acomparative analysis using network simula-tion and optimization techniques to evaluateseveral MANET architectures against a basecase consisting of traditional PTP network tech-nology (i.e., currently fielded technology as offiscal year 2011). We first develop a networkformulation to model key aspects of traditionalPTP and MANET systems. We then simulateand gauge network performance in environ-ments ranging from low-fidelity, theoreticalrepresentations to realistic, high-fidelity com-bat scenarios. This systematic approach en-ables us to examine fundamental propertiesof MANET and build intuition with theoreticaltechniques before using less tractable, higher-fidelity simulations.

We quantify the benefit that MANETs canprovide to tactical communications networksover traditional point-to-point networks. Wealso quantify the value of the use of unmannedaircraft systems (UASs) as airborne nodes inMANETs, a capability especially useful in com-munications scenarios involving rugged terrainand large distances. We find that due to MANETfragility in high-loss environments, small unittactics may need to be modified if commanderswish to fully utilize the advantages offered byMANET communications.

To our knowledge, we are the first to rigor-ously examine and quantify the value of MANETtechnology within the USMC infantry battalion.Our work has been used by the study sponsor,the Marine Corps Tactical Systems SupportAgency (MCTSSA), in improving the fidelityand rigor of MANET test and evaluation efforts.

Previous WorkCoyne et al. (2007) use a human-factors

based assessment to identify the optimal set of

SIMULATION AND ANALYSIS OF MOBILE AD HOC NETWORK TECHNOLOGY IN THE US MARINECORPS INFANTRY BATTALION

0 Military Operations Research, V18 N4 2013

communications gear to support Marines con-ducting distributed operations (DO). They iden-tify a list of information that must be exchangedin order to execute tactical tasks, and methods(visual, auditory, etc.) of exchanging that in-formation. Their recommendations are basedsolely on subject matter expert (SME) input;they do not conduct any simulation or testingto validate their conclusions.

Extensive field testing has been conductedto demonstrate the ability of MANETs to sup-port geographically dispersed tactical combatoperations, including Bommer (2007) and bythe Marine Corps Warfighting Laboratory(Reynolds, 2011). Simmons and Curran (2007)consider the benefits of MANET to a USMC pla-toon operating in a geographically dispersedmanner by examining the results of previousfield testing. Much of this research incorporatessimulation. However, none specifically considersthe impact of MANET technology on each tierwithin the battalion.

Blackshear (2002), Kioumourtzis (2005), andSmith (2009) use computer simulation to com-pare various routing protocols within a MANET,including UAS nodes. Kant et al. (2008) useoptimization to calculate network capacity inMANETs. Karhima et al. (2005) use computersimulation to evaluate MANETs in the presenceof mobile jammers. However, none considernetworks capable of supporting a battalion,nor do they consider the effects of mobile nodesoperating over rugged terrain.

Alderson et al. (2011b) examine the trade-offs inherent in the fielding of MANET radios,specifically the Enhanced Position LocationReporting System (EPLRS). To model networkperformance, they use first principles-basedmodels and discrete event simulation, includ-ing the Simultaneous Routing and ResourceAllocation (SRRA) problem of Xiao et al. (2004).They find that while a small increase in thenumber of nodes does not necessarily have adetrimental impact on network performance,larger networks are more difficult to properlymanage. We also use SRRA to model networkperformance, and our research builds on theirconcept of network dispersion (i.e., a modelof the relative dispersion between nodes).

Much of the research presented in this pa-per was conducted as part of the Reinforced

Infantry Battalion MANET Study, executedby the Marine Corps Combat Development Com-mand (MCCDC) and sponsored by MCTSSA. SeeMCCDC (2011) for the complete study report.

This paper is organized as follows. In thenext section, we describe in detail our tech-niques for modeling tactical communicationsnetworks, and the methods we use to measurethe performance of our models. We describeseveral analyses using these models, and thenbriefly summarize our findings. We concludewith suggestions for follow-on research.

COMMUNICATIONS NETWORKMODEL

Infantry Battalion Network StructureWe create a mathematical network model

to simulate key aspects of an infantry battalioncommunications network. A Marine Corps rein-forced infantry battalion consists of approxi-mately 700 Marines, typically assigned to threeinfantry companies, a weapons company, andother attached units such as an amphibiousvehicle platoon or engineer platoon. Due tosimulation constraints, we model only theleaders at each tier of the battalion. These tiersare the battalion, company, platoon, squad, andfire team, and the respective leaders are the bat-talion commander, company commanders, pla-toon commanders, squad leaders, and fire teamleaders (see Figure 1). Note each tier except thefire team tier has three identical subcomponents;for example, each company has three platoons,each platoon has three squads, and so on (weassume that Weapons Company personnel aredivided among the other battalion elements).As dictated by scenario, we also model air- andseaborne radios and the Fire Support Coordi-nation Center (FSCC) as company-level leaders.Each tier may use different types of radios (andthus use different transmission powers, fre-quencies, etc.), but we assume all nodes at aparticular tier will use the same type of radio.

Modeling Network TopologyWe define each leader within the battalion

as a node within our network model, and define


Military Operations Research, V18 N4 2013 Page 21Military Operations Research, V18 N4 2013 Page 21

N to be the set of all nodes, indexed by i ¼1, 2,., n, where n 5 jNj. Our scenarios modelbetween 40 and 121 nodes, depending on thenumber of personnel and vehicles modeled ineach scenario. Nodes connect via wireless linksor arcs, denoted i; jð Þ 2 A. We distinguish thesepoint-to-point arcs from source-destinationpairs u; vð Þ 2 D � A, which are two nodes withinthe battalion that must exchange communi-cations traffic. In the base case (consisting ofcontemporary PTP technology), no node canserve as an intermediate node, and nodes ex-change traffic only with their immediate seniorsand subordinates (e.g., a company commanderwill only exchange traffic with his battalioncommander and his platoon commanders). Inthe MANET cases, we assume the same source-destination pairs D must communicate, but cer-tain types of radios may serve as intermediatenodes.

A single node may represent one or moreman-packed, vehicular, or airborne radios. How-ever, a node in any arc can connect to the dis-tant node using only one fixed radio (i.e., anode may not dynamically connect using a dif-ferent radio). See Figure 2 for a simple example.Each node (battalion, company, and platoon com-mander) communicates to other nodes usingonly one radio, but the company commander

node communicates using two different radios.We model the characteristics of each radio viadifferent arc capacities (see following sections).

To simulate various network routing con-figurations, we describe several MANET typesby defining different arc sets A. APTP includesonly PTP connections. ADirect allows MANET-capable radios to connect with other MANET-capable radios only in a node’s direct chain ofcommand. ATotal allows all MANET-capable ra-dios to connect, regardless of chain-of-commandassignment. A simple example is presented inFigure 3. A platoon commander must exchangecommunications traffic with the company com-mander. Using APTP, the platoon commander

Figure 1. A simplified infantry battalion hierarchy.Each node except the fire team leader may havethree identical sub-components. Such a battalion hasa total of three companies, nine platoons, 27 squadleaders, and 81 fire team leaders.

Figure 2. Example of the connections between threenodes. Each node may have more than one radio (e.g.,the company commander node), but nodes connectto other nodes using only one radio.

Figure 3. Example of the three routing types, APTP,ADirect, and ATotal.


Page 22 Military Operations Research, V18 N4 2013

can only communicate directly with the com-pany commander (thin black line). With ADirect,the platoon commander can route traffic throughother nodes, such as a squad leader, in his directchain of command (heavy black line). With ATotal,traffic can be routed through any number ofother nodes serving as intermediate nodes(dashed line).

Calculating Network AvailabilityOne of our measures of network perfor-

mance is network availability, defined as thenumber of source-destination pairs u; vð Þ 2 Dthat can successfully exchange traffic. Networkavailability is arguably the most importantmeasure of network performance because with-out it, no nodes can communicate. By simplycounting nodes, we implicitly assume each nodegenerates traffic with the same relative value.This is unrealistic, as the battalion commander’straffic will generally be more important thanany other traffic, and there are instances whena low-tier node could generate very importanttraffic (e.g., while calling in artillery reports).However, objective traffic value data do not ex-ist, and this assumption allows us to obtaingeneralized results.

To calculate network availability, we firstdetermine which nodes i; jð Þ 2 A are connectedwirelessly by calculating the signal-to-noise(SNR) ratio. We assume the environment con-tains white noise that is constant as a functionof time. If the SNR is above the technical thresh-old for the radios in question in both directions,the two nodes are connected. To determine thesignal component of the SNR ratio, we calculatethe RSS rij between each node in dBm usingthe standard link budget formula (Olexa, 2005,p. 79):

rij 5 poweri 1 gi 2 li 2 lpath 2 lmisc 1 gj 2 lj (1)

where poweri is transmitted power in dBm, gi

and gj are respectively the gains of the nodes iand j in dB, li and lj are respectively the losses(i.e., from cables, connectors, etc.) of the nodesin dB, lpath is total path loss in dB, and lmisc ismiscellaneous loss (such as fade margin) in dB.All of the terms in Equation (1) are input data,determined by the equipment and environment,

except for the total path loss lpath, which dependson the physical position of nodes i and j and theintervening terrain.

Our formulation allows the use of anymethod for computing lpath, including the Ir-regular Terrain Model (ITM) (Longley and Rice,1968) and Hata-COST 231 (COST, 1999). We pre-fer the Terrain Integrated Rough Earth Model(TIREM) of Alion Science & Technology Cor-poration (Alion, 2007). This model samplesterrain elevation to compute path loss, and con-siders the effects of free space loss, diffractionaround obstacles, and atmospheric absorptionand reflection.

We assume source-destination pairs u; vð Þ 2D can successfully exchange traffic if a pathconsisting of connected arcs exists from nodeu to node v, and from node v to node u. Todetermine if paths exist between all source-destination pairs u; vð Þ 2 D, we solve the all-pairsshortest path problem using the Floyd-Warshallalgorithm (Floyd, 1962; Warshall, 1962) wherearc costs are inversely proportional to arc RSS.This approach favors both arcs with high RSS,and paths consisting of fewer arcs (or ‘‘hops’’).Our model is similar to actual data networklink state routing protocols, such as the OpenShortest Path First (OSPF) protocol (Moy, 1988;Aggelou, 2005).

Calculating Network ThroughputOur other metric of performance is net-

work throughput, defined as the rate of trafficflow in bits per second (bps) between eachsource-destination pair u; vð Þ 2 D. These pairsdefine the commodities in our multicommod-ity network flow problem (Ahuja et al., 1993).We use the term ‘‘traffic flow’’ to refer generi-cally to the transmission of packets or data-grams during a data session. We adopt andmodify the simultaneous routing and resourceallocation (SRRA) problem of Xiao et al. (2004)to approximate network throughput. First, wecalculate arc capacities (as opposed to arc costscalculated for the network availability metric)between each node using the Shannon capacityformula (1949), which establishes a theoreticalupper bound on transmission capacity in bps.Following Xiao et al. (2004), the capacity fromnode i to node j is:



ðCapacityÞij 5 bandwidth log2 11gainij

noiseijlossijPij

� �

"ði; jÞ 2 A

(2)

where bandwidth is channel bandwidth in Hertz,gainij is the sum of the antilog gain terms (gij andgji), noiseij is the background noise power inwatts or volt2 from node i to node j, and lossij

is the sum of the antilog loss terms (lij, lji, lpath,and lmisc). These are input data calculated basedon the position of the radios within the simula-tion environment. We assume each radio haslimited total transmission power denoted pi (inwatts), and we define Pij to be the amount of pi

used to transmit from i to j. Thus, each node isadditionally constrained byX

j:ði;jÞ2A

Pij # pi: (3)

We wish to measure traffic flow between eachsource-destination pair u; vð Þ 2 D in bps. Wequantify the value of network flow using thelog-utility function of Xiao et al. (2004). Thisfunction more equitably distributes flow thana linear function by assigning diminishingvalue to increasing flow. Note that a zero flowis assigned an infinite penalty, so there is strongincentive to assign some flow to each source-destination pair. Defining Sv

u to be the total floworiginating at node u and destined for node v,we have

ðUtility of Total Network FlowÞ[X

u

Xv6¼u

log2ðSvuÞ: (4)

Our version of the Xiao et al. (2004) SRRA prob-lem follows:

Formulation SRRAIndex Use

i 2 N node (alias j, k)i; jð Þ 2 A directed arc (link)uð Þ 2 D � N destination (alias v)

Input Datagainij sum of antilog gain terms from i 2 N to j 2 N [none]lossij sum of antilog loss terms from i 2 N to j 2 N [none]pi maximum total transmission power per node [watts]bandwidth channel bandwidth [hertz]noiseij background noise power from i 2 N to j 2 N [watts]

Decision VariablesSv

u total flow of traffic from u 2 N to destination v 2 D [bps]Fv

ij traffic flow along arc i; jð Þ 2 A to destination v 2 D [bps]Tij total flow along arc i; jð Þ 2 A [bps]Pij total transmission power along arc i; jð Þ 2 A [watts]

Formulation

maxS;F;T;P

Xu

Xv6¼u

log2ðSvuÞ (S0)

s:t:X

k:ð j;kÞ2A

Fvjk 2

Xi:ði;jÞ2A

Fvij 5 S

vj "j 2 N;"v 2 D (S1)

Tij 5X

d

Fvij "ði; jÞ 2 A (S2)



Given fixed node locations, this is a multi-commodity network flow problem. The objec-tive function (S0) maximizes the total utilityof traffic flow between each source-destinationpair. Constraints (S1) ensure balance of flowat each node. Constraints (S2) define the totalflow along any arc as the sum of all traffic flowsalong that arc. Constraints (S3) ensure that totalflow along each arc is less than or equal to itscapacity. Constraints (S4) ensure that total trans-mission power at each node is conserved. Con-straints (S5)-(S8) ensure nonnegativity.

As observed by Xiao et al. (2004), the SRRAproblem has special structure that allows it tobe solved via dual decomposition. Specifically,the dual function can be evaluated separatelyin the network flow variables S, F, T, and thecommunications variable P. Xiao et al. (2004) ob-serve that the dual function is always convex.They assume that Slater’s condition holds (seeBoyd and Vandenberghe, 2004, Sec. 5.2)—thatis, a feasible solution (S,F,T,P) exists such thatthe nonlinear capacity constraints hold withstrict inequality. They conclude that strong dual-ity holds and the solution to the primal problemis equal to the solution of the dual problem.

However, Xiao et al. (2004) also note thatthe objective function of the primal problem isnot strictly concave in the variables F and T,and thus the dual function is only piecewisedifferentiable. As a result, the dual problem is anondifferentiable convex optimization problem,to which they apply the subgradient methodto obtain a solution. Each iteration of the sub-gradient method might not necessarily improvethe dual objective value, but each iteration reduces

the distance to the optimal solution (Bertsekas,1999, p. 621). See Xiao et al. (2004), Shankar(2008), and Nicholas and Alderson (2013) fordetails on solving the SRRA problem.

We use the General Algebraic ModelingSystem (GAMS, 2011) and the CONOPT (2011)solver to solve the SRRA problem to near opti-mality. We use total delivered flow betweeneach source destination pair Sv

u as a measureof network throughput. We calculate both net-work availability and network flow at eachtime step within the simulation. Our modelsof arc connections, source-destination paths,and throughput are useful simplifications ofhow actual radio links are established and main-tained. These first-principles, physics-basedmodels establish a performance upper bound,because ultimately no modulation scheme,waveform, protocol, or other characteristic canovercome the physical-layer limitations of radiophysics.

ANALYSIS AND RESULTSWe use GAMS (GAMS, 2011), Microsoft

Visual Basic for Applications (VBA) (Microsoft,2011) and Analytical Graphics, Inc. (AGI) Satel-lite Toolkit (STK) (AGI, 2011) to simulate com-munications network topologies within bothgeneric battalion formations on flat terrain, andwithin our realistic tactical scenarios. The threescenarios respectively model an amphibiousassault, a mechanized movement to contact,and an irregular warfare operation, and includedetailed node movement plans and terrain

Tij 2 bandwidth log2 1 1gainij

noiseijlossijPij

� �# 0 "ði; jÞ 2 A (S3)

Xj:ði;jÞ2A

Pij # pi "i 2 N (S4)

Svu $ 0 u 6¼ v (S5)

Fvij $ 0 "ði; jÞ 2 A;"v 2 D (S6)

Tij $ 0 "ði; jÞ 2 A (S7)

Pij $ 0 "ði; jÞ 2 A (S8)



information. See MCCDC (2011) and MajorCombat Operation21 (2007) for complete de-tails on our scenarios.

We conduct deterministic, simulation-basedcomparative analysis between the base case andseveral MANET cases. We model both theoreticalradios and actual fielded and planned radiosets using technical specifications providedby the radio manufacturers and other subjectmatter experts. The base case consists of non-MANET radios operating in the low VHF range(65 MHz). In this paper, we consider only twoMANET cases; they are based on actual radiosbut we withhold their nomenclature and exactspecifications. The first (VHF MANET) con-sists of radios operating in the high VHF range(225 MHz), and the second (UHF MANET)consists of radios operating in the UHF range(1.125 GHz). See MCCDC (2011) for completedetails on all of comparative cases.

Using network availability and throughputas a proxy for network performance, we mea-sure network performance both overall and asa series of time-stepped instances at 1-minuteintervals during our scenarios. We examine per-formance both for each tier of the infantry bat-talion (battalion, company, platoon, and squad),and for the entire battalion, on both theoreticalflat terrain and across all three tactical scenarios,and using both theoretical and actual radio spec-ifications. For the sake of brevity, the followingsub-sections present our summarized analytic

results by major findings. See MCCDC (2011)for complete results.

The Effect of the Ability to RouteTraffic

Our first analysis considers the effects ofthe ability to route traffic on network avail-ability. We model theoretical radios operatingat 65 MHz and 10 watts of transmission power,and examine network availability for routingtypes APTP, ADirect, and ATotal. Figure 4 presentsthe battalion network availability for the am-phibious assault scenario as a function of time,results typical of all three scenarios.

In this scenario, a total of 26 landing craft,aircraft, and other vehicles transiting ashorecommunicate with the battalion commander lo-cated on a ship. The vertical axis depicts thenumber of nodes that are able to communicatewith the battalion commander (i.e., networkavailability), and the horizontal axis depictsscenario time. The effects of the various as-sault waves are evident in the ‘‘No Mesh’’ linein Figure 4: the vehicles gradually move ashoreup to time 05:30, then return to the ship at06:00, and again go to the shore around 06:15.

We calculate average availability by aver-aging network availability at each time stepover the course of the scenario. Without theability to route traffic, the point-to-point net-work delivers an average battalion network

Figure 4. Battalion network availability during the amphibious assault scenario for three routing types.



availability of 43 percent—that is, on average43 percent of the company-level units can suc-cessfully exchange traffic with the battalioncommander. Networks with the ability to routetraffic (routing types ADirect and ATotal) providemuch greater average network availability of67 and 75 percent, respectively, because theyare able to use intermediate nodes to route traf-fic between the ship and vehicles ashore. Notethere are moments when the routable networksdo not provide greater network availability; thisoccurs when there are limited or no intermediatenodes available for routing. Also, near the end ofthe scenario network availability does not go tozero because some of the assault vehicles returnto the ship.

Figure 5 presents battalion network avail-ability for all three scenarios, averaged over timefor the three routing types. As with the am-phibious assault scenario, we find that all otherthings being equal, the ability to route trafficcan greatly increase network availability.

The Effect of Higher FrequenciesHowever, all other things are not typically

equal. Perhaps the most significant physical-layer difference is that MANET radios gener-ally operate at higher frequencies than PTPradios. These higher frequencies are subject togreater signal attenuation than lower frequen-cies (Singal, 2010). We next explore the abilityof MANET radios to overcome this inherent

disadvantage by routing traffic. We model the-oretical APTP radios operating at 65 MHz, andMANET ATotal radios operating at both 200 MHzand 400 MHz. Figure 6 presents average battal-ion network availability for all three scenarios.

In the amphibious assault scenario, the abil-ity to route traffic does not overcome the rangelimitations of higher frequencies, and neitherroutable network provides availability equal tothe PTP network at 65 MHz. However, in the ir-regular warfare scenario, both the routable net-works provide greater network availability thanthe PTP network. The 400 MHz system actuallyprovides greater availability than the 200 MHzsystem, due to the nonlinear effects of terrainon radio wave propagation. We find that theability to route traffic alone may not be suffi-cient to overcome the inherent range limita-tions of higher frequencies.

The Effect of Greater TransmissionPower

We next consider the effect of greater radiotransmission power to overcome the range lim-itations of higher frequencies. We use our theo-retical radio models to compare a theoreticalAPTP radio operating at 65 MHz with 10 wattsof transmission power, to ATotal routable sys-tems operating at 400 MHz with transmissionpowers ranging from 10 to 100 watts. Figure 7presents average battalion network availabilityfor all three scenarios.

Figure 6. Average battalion network availabilitywith varying operating frequencies.

Figure 5. Average battalion network availability forthree routing types. All other things being equal, in-creased connectivity provided by the ability to routetraffic results in greater network availability.



For all three scenarios, greater transmissionpower provides greater availability for routablenetworks, results similar to Alderson et al.(2011b). Note that in the amphibious assaultscenario, a substantial amount of transmissionpower (100 watts) is required for a routablenetwork at 400 MHz to provide the same levelof availability as a PTP network operating at65 MHz with a mere 10 watts of transmissionpower. Per Equation (1), received signal strength(and thus network availability) can be increasedby increasing antenna gain or transmissionpower, or by decreasing loss. Greater transmis-sion power and/or more sensitive receiversand antennae may be necessary for MANET

radios to provide the same network availabilityas non-MANET radios at lower frequencies.

The Effect of Intermediate NodesThe remainder of our analysis considers

only actual, real-world radios. We next considerthe effect of the presence of intermediate nodeson MANET performance, comparing the PTPbase case against the VHF and UHF MANETcases. See MCCDC (2011) for a detailed list ofspecifications. Figure 8 presents the battalionnetwork results for the movement-to-contact(MTC) scenario, results representative of allscenarios and network tiers. Note this scenario

Figure 7. Average battalion network availability with varying transmission powers.

Figure 8. Battalion network availability for movement-to-contact scenario.



includes an FSCC, so there can be a maximumof four nodes connected to the battalion node.

We see that over the entire scenario, theMANET cases deliver higher average battalionnetwork availability than the PTP base case,but there is a substantial period of time (01:44-04:48) in which neither MANET case can pro-vide the same network availability as the basecase. Similar outages occur in all three scenar-ios and at different tiers within the battalion.These outages occur when intermediate nodesare unable to relay traffic from source to desti-nation. In this scenario, the company nodes(mounted in armored vehicles) transit behindterrain that blocks RF transmission, and noother nodes are able to relay the required com-pany traffic to the battalion commander node.MANETs can provide increased network avail-ability only when intermediate nodes are inrange, yet the location of nodes is often drivenby operational (rather than communications) re-quirements. Without these nodes, a MANET isessentially reduced to a PTP network, and asMANETs operate at higher frequencies (subjectto greater signal attenuation), a MANET forcedinto PTP mode will provide less network avail-ability than a traditional, lower-frequency PTPnetwork.

The Effect of Dedicated AirborneRelay Nodes

We next consider the benefit of a dedicatedairborne relay node, such as an unmanned air-craft system (UAS). Dedicated airborne relay

nodes are often used to increase network per-formance (Krout et al., 2003) because radiowaves transmitted from airborne nodes aresubject to much less signal attenuation fromterrain features than terrestrial nodes, and thuscan generally connect at much greater distances.We compare our PTP base case against bothMANET cases and a case with an added UASnode (modeled as a company-level node) toserve as a dedicated airborne relay node forbattalion traffic. This device serves as neithera source nor destination for any communica-tions traffic, but solely as a relay. Figure 9 pres-ents the average battalion network availabilityresults for all three scenarios. The left side pres-ents the base case and the VHF MANET case,and the right side presents the base case andthe UHF MANET case.

In all three scenarios, the addition of a UASto a MANET provides network availability atleast equal to that provided by the originalMANET. In all but one scenario (amphibiousassault with UHF MANET radios), the UAS in-creases network availability. Note that in theirregular warfare scenario, both the VHF andUHF MANETs are unable to provide networkavailability equal to the base case without theaddition of the UAS, due to the impact of ter-rain on higher frequencies. In the amphibiousassault scenario, even the addition of a UASis unable to overcome the inherent range limi-tations of UHF MANET. We find that althoughdedicated airborne relay nodes are not a panaceafor availability shortfalls, they can greatly in-crease network availability, and may be required

Figure 9. The effect of the addition of a dedicated airborne relay on battalion network availability for eachscenario.



in certain circumstances for MANETs to pro-vide the same level of network availability asa lower-frequency PTP network.

The Effect of Additional LossThus far, we have examined network avail-

ability with the assumption that signal loss(other than that due to terrain obstructions) isconstant. Per Equation (1), additional signalloss will reduce the SNR and may reduce net-work availability. We now explore the effectof additional loss on network availability. Thisadditional loss could represent active jamming,background noise, vegetation, inclement weather,bad connectors, and so on. We incrementally addup to 30 dB of additional loss to each point-to-point arc (i, j) 2 A, and we assume this loss isconstant across the entire spectrum. We con-duct this analysis across all scenarios and cases.Figure 10 displays the results for the movement-to-contact scenario, representative of the othertwo scenarios.

In this chart, the horizontal axis indicatesthe additional signal loss introduced to allarcs, from 0–30 dB, and the vertical axis indi-cates average battalion network availabilityas a percentage. These results indicate thatin high-loss environments, the base case mayprovide greater network availability than theMANET cases. The following simple exampleexplains these results.

Consider two nodes that must communi-cate. In a point-to-point network, these twonodes must communicate directly. In a MANET,the nodes are able to use intermediate nodes torelay traffic. A point-to-point arc has surplusreceived signal strength if the received signalstrength is greater than the required threshold.This is analogous to having extra range.

In the scenarios and cases considered inthis study, the base case arcs (i.e., VHF point-to-point arcs) typically have greater surplus re-ceived signal strength than the MANET arcsdue primarily to the lower transmission fre-quency of the former. Hence, in environmentswith greater loss, those point-to-point arcs inMANETs (with less or no surplus signal strength)will disconnect first. A MANET is capable ofrerouting traffic along alternate paths, if theyexist. However, our conceptual radio modelcalculates the optimal path between any twonodes, so any other path will have connectionswith even less surplus received signal strengthand/or more hops than the selected optimum.

In the realistic scenarios considered in thisstudy (and by Hong et al., 1999), nodes tend toremain relatively close to other nodes in theirimmediate chain-of-command (e.g., squad nodeswithin a particular platoon remain near eachother). This ‘‘clumping’’ prevents the routing oftraffic because possible intermediate nodes areinsufficiently dispersed. Should the intermedi-ate nodes be more evenly dispersed, traffic can

Figure 10. The effect of additional signal loss on battalion network availability in the movement-to-contactscenario.



be routed between the required nodes. Theseresults suggest that rather than MANET en-abling dispersion, dispersion may enable MANET.In other words, in a high-loss environment,MANETs may not fully function without suf-ficient dispersion.

This finding has a powerful implication.Traditionally, communications networks sup-port military operations, not the other wayaround. Certainly, the ability to communicatehas always been limited by technical constraints,and MANET technology is not a panacea to thisproblem. MANEToffers the military commanderexciting new capabilities, but if he or shewishes to utilize the full capabilities of MANET,including extending MANET connectivity downto the lowest levels, then tactics may need to beadjusted to support this requirement.

The Effect of Channel Bandwidthon Network Throughput

According to Equation (2), higher frequencysystems (generally subject to greater signal

attenuation) require greater bandwidth to pro-vide throughput equal to systems operating atlower frequencies. Our final analysis considersthe allocation of bandwidth to enable MANETsystems to provide sufficient throughput rates.To calculate total network throughput, we solveour version of the SRRA problem using GAMS(2011) at each time step within our scenarios,and sum the throughput values between eachsource-destination pair u; vð Þ 2 D at each tierin the battalion.

First, we examine our three cases using the-oretical radios operating at the same band-width (500 KHz). The results for the battalionnetwork in the MTC scenario are presented inFigure 11. The vertical axis indicates the totalnetwork throughput between each companynode and the battalion node. Even without theability to route traffic (i.e., a PTP network), the65 MHz PTP system is able to provide sub-stantially more network throughput than the225 MHz or 1.125 GHz MANET system. Thisis in keeping with our earlier analysis show-ing the propagation limitations of higher

Figure 11. Network throughput for theoretical radios using a 500 kHz channel. By adding the ability to route,higher-frequency systems can outperform lower-frequency systems without the ability to route.



frequencies. By simply adding the ability toroute (i.e., a MANET), the higher-frequencysystems are able to outperform the 65 MHzPTP system. We find that the ability to routetraffic alone can increase network through-put; larger channel bandwidths may not benecessary.

Finally, we measure network throughputwhile varying both operating frequency andbandwidth for a MANET system in the MTCscenario. The results for the battalion networkare presented in Figure 12. In this chart, the areaof each point is proportional to the sum of theaverage throughputs between the company andbattalion nodes. Such a plot serves as a visualaid in understanding the effects of frequencyand bandwidth on network throughput. Here,doubling the bandwidth doubles networkthroughput, as predicted by Equation (2). How-ever, halving the operating frequency resultsin only a marginal increase in network through-put, due to the specific (nonlinear) effects ofterrain on RF propagation. Such informationcan inform RF resource allocation decisions.

CONCLUSION ANDRECOMMENDATIONSFOR FUTURE WORK

The Marine Corps is actively pursuing theuse of MANET technology to extend digitalcommunications throughout the battlefield.Yet, MANET technologies are not a panaceato tactical communications difficulties, nor doestheir use alone enable greater dispersion offorces. The requirement for sufficient inter-mediate nodes implies that greater fieldingof MANET radios to all echelons may increasenetwork availability throughout the infantrybattalion. As radio technology continues toapproach limitations imposed by physics, andUSMC forces increasingly operate in electro-magnetically noisy urban environments, dedi-cated airborne relay nodes—immune to thesignificant propagation losses suffered by terres-trial radios—will become increasingly impor-tant. The sensitivity of MANETs to high-lossenvironments implies that if the USMC is topush data connectivity down to the lowest

Figure 12. Network throughput for theoretical MANET radios, varying both bandwidth and frequency. Thearea of each point is proportional to the network throughput, ranging from 8.47 Mbps to 3.88 Gbps.



levels within the battalion, tactics may need tobe modified to support the full use of MANETs.The ability to route traffic alone can increasenetwork throughput, making MANETs par-ticularly useful in bandwidth-constrained en-vironments.

Possible future work could include anexamination of specific network disruptions(accidental or intentional) on the design andoperation of MANETs (e.g., Grotschel et al.,1995; Shankar, 2008; Alderson et al., 2011a),or the effect of slight perturbations of node lo-cations. This research assumes sufficient EMspectrum exists; ongoing research at MCCDCconsiders the efficient allocation of spectrumamong several MANET systems. Additionally,our first principles-based analysis techniquecould complement bench-top and field testing,to support model-test-model research.

ACKNOWLEDGEMENTSThe authors are indebted to Drs. David

Alderson and Emily Craparo, and Mr. RexBuddenberg of the Naval Postgraduate School(NPS) for technical support and assistance dur-ing the course of this research. The authors alsothank several colleagues for constructive com-ments on this and earlier versions of this paper.

ABBREVIATIONS AND ACRONYMSAA: amphibious assault scenarioAGI: Analytical Graphics IncorporatedAP: access pointbps: bits per secondC2: command and controlCOTM: communications-on-the-moveDO: distributed operationsDR: disaster reliefEPLRS: Enhanced Position Location Report-

ing SystemGHz: gigahertzGUI: graphical user interfaceHA: humanitarian assistanceISR: intelligence, surveillance, and recon-

naissanceIW: irregular warfare scenariokHz: kilo HertzITM: Irregular Terrain Model

kbps: kilobits per second (1,000 bits persecond)

LOS: line of siteMANET: mobile ad hoc networkMbps: megabits per second (1,000,000 bits

per second)MCCDC: Marine Corps Combat Develop-

ment CommandMCTSSA: Marine Corps Tactical Systems

Support ActivityMHz: mega HertzMTC: movement-to-contact scenarioPTP: point-to-pointRSS: received signal strengthSME: subject matter expertSNR: signal-to-noise ratioSRRA: Simultaneous Routing and Resource

AllocationSTK: Satellite ToolkitTIREM: Terrain Integrated Rough Earth

ModelUHF: ultra high frequencyUSMC: United States Marine CorpsVHF: very high frequencyWMN: wireless mesh network

REFERENCESAggelou, G. 2005. Mobile Ad hoc Networks: From

Wireless LANs to 4G Networks, McGraw-Hill.

Ahuja, R., Magnanti, T., and Orlin, J. 1993.Network Flows: Theory, Algorithms, andApplications, Prentice-Hall.

Alderson, D.L., Brown, G.G., Carlyle, W.M., andWood, R.K. 2011. Solving Defender-Attacker-Defender Models for Infrastructure Defense,Proc. 12th INFORMS Computing SocietyConference, INFORMS, pp. 28–49.

Alderson, D.L., Craparo, E.M., and Fry, W.J.2011. Assessing Tradeoffs in Mobile Ad hocNetwork Deployment: A Case Study in GroundSoldier Mobile Systems, Technical Report01-2011, Naval Postgraduate School.

Analytical Graphics, Inc. 2011. STK. http://www.agi.com/products/.

Alion Science and Technology Corporation.2007. TIREM Details. http://www.alionscience.com/index.cfm?fuseaction¼Products.viewpage&productid¼19&pageid¼34.



Bertsekas, D. 1999. Nonlinear Programming,Athena Scientific.

Blackshear, H. 2002. Developing a ConceptualUnmanned Aerial Vehicle CommunicationsMobile Ad Hoc Network Simulation Model,MS thesis, Systems Technology, NavalPostgraduate School.

Bommer, J. 2007. Joint Networking Commandand Control (C2) Communications amongDistributed Operations, JCAS, and Joint Fires,MS thesis, Systems Technology, NavalPostgraduate School.

Boyd, S., and Vandenberghe, L. 2004. ConvexOptimization, Cambridge University Press.

A Concept for Enhanced Company Operations,Headquarters, US Marine Corps, August 28,2008.

CONOPT. 2011. ARKI Consulting and Devel-opment, Denmark.

Corson, S., and Macker, J. 1999. Mobile Ad hocNetworking (MANET): Routing ProtocolPerformance Issues and Evaluation Consider-ations, IETF RFC 2501; http://www.ietf.org/rfc/rfc2501.txt.

COST (European Cooperation in the Field ofScientific and Technical Research). 1999.Digital Mobile Radio Towards Future GenerationSystems, COST 231 Final Report.

Coyne, J. T., Stripling, R., Pfluger, K. C.,La Budde, Z., and Afergan, D. 2007.Company and Below Command andControl Information Exchange Study,Naval Center for Applied Research inArtificial Intelligence, Naval ResearchLaboratory.

Floyd, R. W. 1962. Algorithm 97, Communicationsof the ACM, Vol 5, No 6, p. 345.

GAMS Development Corp. 2008. GeneralAlgebraic Modeling System (GAMS),http://www.gams.com.

Goulding, V. J. 2009. Enhanced MAGTFOperations: Capitalizing on Lessons Learned,Marine Corps Gazette, 93-8, pp. 13–21.

Grotschel, M., Monma, C. L., Stoer, M. 1995.Design of Survivable Networks, NetworkModels, M. O. Ball, T. L. Magnanti,C. L. Monma, and G. L. Nemhauser, eds.Handbooks in Operations Research andManagement Science, Vol 7. ElsevierNorth-Holland, pp. 617–669.

Harris Corp. 2011. Harris Corporation Receives$25 Million in Orders from US Marine Cropsfor Falcon III AN/PRC-117G MultibandRadio Systems, http://rf.harris.com/news/view_pressrelease.asp?act¼lookup&pr_id¼3280.

Hong, X., Gerla, M., Pei, G., and Chiang, C. 1999.A Group Mobility Model for Ad Hoc WirelessNetworks, Proceedings of the 2nd ACM Inter-national Workshop on Modeling, Analysis, andSimulation of Wireless and Mobile Systems.

Kant, L., van den Berg, E., and Krishnan, K.2008. Mobile Ad hoc CommunicationNetworks: Capacity Analysis and Design.Proceedings 4th International Conference onWireless and Mobile Communications, IEEE,pp. 84–90.

Karhima, T., Lindroos, P., Hall, M., andHaggman, S. 2005. A Link Level Study of801.11b Mobile Ad hoc Network in MilitaryEnvironment, Proceedings of the IEEE MilitaryCommunications Conference, IEEE, Vol. 3,pp. 1883–1186.

Kioumourtzis, G. 2005. Simulation and Evalua-tion of Routing Protocols for Mobile Ad HocNetworks (MANETs), MS thesis, SystemsEngineering and Computer Science, NavalPostgraduate School.

Krout, T., Durbano, S., and Shearer, R. 2003.Network-centric Applications and TacticalNetworks, Eighth International Command andControl Research and Technology Symposium.

Longley, A. G., and Rice, P. L. 1968. Predictionof Tropospheric Radio Transmission Loss OverIrregular Terrain. A Computer Method-1968,Institute for Telecommunications Sciences.

Major Combat Operation – 1, Swiftly Defeat theEfforts 2014 Multi-Service Force Deployment.2007. Department of Defense Analytic Agenda.

Marine Corps Combat Development Command(MCCDC). 2011. Reinforced Infantry Battal-ion Mobile Ad-hoc Network (MANET) Study,United States Marine Corps.

Marine Corps Systems Command. 2011.Tactical Communications Systems,http://www.marcorsyscom.usmc.mil/sites/cins/TCS/index.html.

Microsoft. 2011. Visual Basic DevelopmentCenter, http://msdn.microsoft.com/en-us/vbasic/ms789056.



Moy, J. 1998. OSPF Version 2. IETF RFC 2328,http://www.ietf.org/rfc/rfc2328.txt.

Nicholas, P., and Alderson, D. 2013. Fast,Effective Transmitter Placement in WirelessMesh Networks, Military Operations Research,Vol 17, No 4, pp. 69–84.

Olexa, R. 2005. Implementing 802.11, 802.16, and802.20 Wireless Networks: Planning, Trouble-shooting and Operations, Elsevier.

Reynolds, J. 2011. Aerial Mobile Ad-Hoc Net-work (MANET) Passive Relay (AMPR) LTA 1,Marine Corps Warfighting Laboratory.

Shankar, A. 2008. Optimal Jammer Placement toInterdict Wireless Network Services, MSthesis, Operations Research, Naval Post-graduate School.

Shannon, C. 1949. Communication in thePresence of Noise, Proceedings of the IRE,Vol 37, pp. 10–21.

Simmons, R., and Curran, C. 2007. ‘‘MeshNetworks within a Distributed Operations

Framework Utilizing IP Based Radios,’’ MSthesis, Information Technology Management,Naval Postgraduate School.

Singal, T. L. 2010. Wireless Communications, TataMcGraw Hill.

Smith, H. D. 2009. Simulation-Based Analysisof Tactical Multi-Hop Radio Networks, MSthesis, Computer Science, Naval PostgraduateSchool.

United States Marine Corps. 2008. MarineCorps Vision and Strategy 2025.

Warshall, S. 1962. A Theorem on BooleanMatrices, Journal of the ACM, Vol 9, No 1,pp. 11–12.

Xiao, L., Johansson, M., and Boyd, S. 2004.Simultaneous Routing and Resource Alloca-tion Via Dual Decomposition, IEEE Transactionson Communications, Vol 52, No 7, pp. 1136–1144.

Zhang, Y., Luo, J., and Hu, H. 2006. WirelessMesh Networking: Architectures, Protocols, andStandards, Auerbach Publications.



Simulation and Analysis of Mobile Ad-hoc Network Technology in...

Documents

Transcript of Simulation and Analysis of Mobile Ad-hoc Network Technology in...