IWS - The Information Warfare Site
News Watch Make a  donation to IWS - The Information Warfare Site Use it for navigation in case java scripts are disabled

Bandwidth Supply and Demand in Today's Army

The Office of the Joint Chiefs of Staff has asserted, in its 2001 publication Joint Vision 2020, that "[i]nformation, information processing, and communications networks are at the core of every military activity." In response to that guidance, the Army is seeking to create an information technology (IT) environment in which computers are routine battlefield tools, information (measured in bits per second) passes continuously over wired and wireless channels, and the complex communications infrastructure necessary to keep all of that gear operating is in place. This Congressional Budget Office (CBO) study focuses primarily on how the Army transmits streams of information, known as bandwidth, and how much information it wants to transmit, both now and in the future. (Box 1 defines and discusses several IT terms and concepts.) In particular, the study addresses mismatches between the amount of bandwidth demanded at various command levels and the amount of bandwidth supplied.
Box 1.
Information Technology: Some Concepts and Definitions

To allow a better understanding of the analysis and conclusions presented in this Congressional Budget Office (CBO) study, some common terms and concepts from this technical field are discussed below.

Measuring Information Flow
The rate of flow of information is usually expressed inbits per second, a bit being the smallest representation of information in a binary code of zero or one. A shorter phrase for bits per second is bandwidth, as defined in the applied communications industry.1 This report uses bandwidth synonymously with the term throughput.

The more bits per second that an information flow contains, the higher the bandwidth. Convenient units for bandwidth today are typically kilobits per second (thousands of bits per second, or Kbps), megabits per second (millions of bits per second, or Mbps), or gigabits per second (Gbps), which is a thousand times more. Most communications equipment today operates at rates ranging from a few kilobits per second to one or more gigabits per second.

Voice Transmission Rates. For the spoken word, actual information content flows at hundreds to a few thousands of bits per second. For example, the phrase "There's a fire in the room out back" is usually spoken in about a second. That is, in a second, 35 characters, each represented by eight bits of information (in the ASCII code, for instance), or some 280 total bits, are communicated. Very rapid speakers, such as auctioneers, might convey 3,000 bits of information per second.

When voices are transmitted over a phone system today, much more than just the core information is conveyed; the transmission also includes tone, volume, timbre, and other nuances that distinguish one voice from another. After a century of engineering, the number of those additional context bits per bit of core information is about 10 times more. Consequently, commercial phone systems do not operate at 3 Kbps but are instead usually rated at a minimum of 32 Kbps per phone line. The Army standard is somewhat lower--16 Kbps per phone line--but is essentially as satisfactory for most users.

Data Transmission Rates. Although people who talk on the phone are content with the level of communications provided by 32 Kbps of bandwidth, they demand much higher throughputs when using their home computers on the Internet. Today, many people who access the Internet over a phone line use a 56 Kbps modem. (A modem is a device that converts signals produced by, say, a computer to a form compatible with, for example, a telephone.) Fifteen years ago, a typical speed for a modem was 1,200 bits per second. As modem speeds began doubling, and doubling again, 56 Kbps was the first modem speed that took full advantage of the rated throughputs of standard phone lines and telephone interfaces--as well as nearly accommodating the first level of improvement above 56 Kbps, which is usually 64 Kbps.2 (Faster speeds require alternative technology, such as a digital subscriber line, or DSL, or a cable system that bypasses the phone system's copper wires.)

Video Transmission Rates. As was the case with the above analysis of phone lines, it is relatively simple to crudely estimate the size of the bandwidth required for transmitting video images. The typical computer screen today is about 1,000 pixels across by 1,000 pixels high. (Pixels are the small, individually modifiable elements of a video screen.) Basic color representation requires a minimum of eight bits of information per pixel. The cinematic illusion of movement requires about 32 frames per second.3

In addition, just as the phone system requires an extra order of magnitude of bandwidth for the context bits, so, too, computer systems require context bits in about the same 10-to-1 ratio. (The context bits in video transmissions, however, handle such features as noise, packet switching, error correction, and retransmission.) Therefore, if no adjustments were made, the bandwidth required for a single continuous video transmission (sometimes called streaming video) would be about 2,560 Mbps [2,560 bps = (1,000 x 1,000 pixels/frame) x (8 bits/pixel) x (32 frames/second) x (10 context bits/bit)].

Fortunately, modern engineering practices reduce that information load by about a factor of 100 by applying data compression and so-called delta techniques.4 Moreover, trimming the number of pixels in the frame slightly and the frames per second somewhat can reduce the load further, by a factor of 10. Therefore, with a penalty of only a slight loss in resolution and an increase in flickering, 2,560 Mbps can be reduced to a bandwidth load of just a few megabits per second, which is typical for a single point-to-point video transmission.

Software Layers
In a computer network, the software necessary for its operation is sometimes differentiated into layers, and measurements of information bandwidth differ from layer to layer. An e-mail system is an example of applications-layer software. If an output--that is, an e-mail--is to be sent to another user across a large network, the outputs of the e-mail software must be augmented by information from local operating systems, network software, and numerous other software layers. If at some stage in the transmission a radio is employed, more software (the data link and physical layers) is needed at the lowest network layers to transmit bits over an antenna, monitor whether the bits are transmitted without interference, and initiate retransmissions if errors occur. A higher level of bandwidth is required at the data link and physical layers than the level needed to support a throughput at the applications layer.

This Congressional Budget Office study distinguishes between the two levels. Bandwidth associated with the data link and physical layers (the throughput that is often cited by hardware manufacturers) is called engineering bandwidth; bandwidth at the applications layer is referred to as effective, oroperational, bandwidth. Specifically, CBO's analysis compares the demand for and supply of effective (or operational) bandwidth for the Army's point-to-point communications, because that level of throughput is what personnel observe in the field.

1.  That definition of bandwidth is not the one used in scientific fields or in much of the engineering community. There, bandwidth is defined as the difference between two frequencies (radio frequencies, for instance) rather than bits per second. However, the two definitions can be functionally related and are nearly proportional at low noise levels. Thus, communications specialists often shift back and forth, without confusion, between the two definitions in different technical contexts.
2.  The actual throughput of a typical copper point-to-point commercial phone line varies and depends not only on commercial switching rates but on the quality of the telephone as well. Although most phone lines are conservatively rated at 32 Kbps, they can operate easily at 64 Kbps when they have been equipped with higher-quality interfaces.
3.  Some observers notice "jumpiness," or monitor flicker, if the speed of motion is much less than that. Commercial standards today are usually better, or about 60 frames per second.
4.  Delta techniques are a set of algorithms for determining those portions of the new frame that are different from the last frame. Once determined, only those different portions are transmitted.

The Army faces a number of problems in implementing its IT strategy on the battlefield. The service needs much more bandwidth than it has available today to support both its current systems and those planned for the future, but the cost of additional bandwidth is high. Moreover, as this study concludes, the Army's planned investments in communications capability will not fulfill its projected future requirements. Those conclusions have been echoed in numerous discussions that CBO has conducted with members of the Army's IT community as well as in eight recent studies on various aspects of the service's demand for bandwidth.(1) None of those studies, however, has attempted to systematically assess the Army's total bandwidth requirements by communications network and command level.

In this study, CBO analyzes the current and future total demand for communications bandwidth to support operations officers at all tactical levels of command (from the corps down to the individual squad or vehicle) within the Army. The analysis then compares that demand with the total bandwidth supplied by the service's current and planned communications systems. This chapter considers bandwidth supply and demand today; Chapter 2 estimates supply and demand in 2010. Chapter 3 discusses several options for addressing the bandwidth supply/demand gap that exists today and is expected to persist--although at a different point in the command structure--in the future.

Background on the Army's Digitization Initiative

The Army's battlefield communications system was originally designed to handle voice and character-based (old-fashioned terminal teletype) messages. In general, those types of communications, whose upper limit on rate of flow is several kilobits per second, require relatively little bandwidth. To send 100 poor- to average-quality voice messages simultaneously, therefore, requires bandwidth in the range of several hundred kilobits per second. Rates for character-based information are similar. The difference between several and several hundred kilobits per second generally defines the range in bandwidth, or throughput, of the tactical radios that the Army employs today on the battlefield. (Certain augmentations to that equipment were fielded during the recent Iraq war and constitute exceptions to that statement. They are discussed at the end of the chapter.) Using mobile retransmission towers (relays) around the battlefield and the current generation of satellite communications, the Army can meet its current needs for bandwidth for voice and character-based communications.

To transmit computer data and video signals, however, requires much more throughput. Such transmissions on today's battlefields and those of the future include video "downlinks" from unmanned aerial vehicles (UAVs), computer displays of data depicting battlefield situations, video teleconferencing, and telemedicine, in which doctors examine wounds and advise on treatment remotely. Much of the bandwidth demand reflected in those examples is associated with streaming (continuous) video imaging, which requires a minimum of several megabits per second of throughput to be considered continuous by most viewers.

If 100 video and data transmissions are to be supported simultaneously--a typical scenario in a larger headquarters such as for a division or corps, the demand for bandwidth grows from several megabits per second to several hundred. That throughput is three orders of magnitude (10 x 10 x 10, or 1,000, times) greater than the Army's current communications system was originally designed to handle. It is more than 10 times larger than the demand that the system is projected to be able to handle in the future, following the improvements that the Army is currently planning.

The substantial increase in bandwidth demand on today's battlefield is due to the Army's digitization initiative. Begun in 1992 by then Army Chief of Staff General Gordon Sullivan, the initiative was designed to field advanced information technologies (computers, software, databases, and communications equipment) to rapidly provide "situation awareness" and support for decisionmaking. The phrase situation awareness denotes knowing the locations of both U.S. forces (in military terms, referred to as Blue situation awareness) and enemy troops (known as Red situation awareness). Within a short time, Army leaders had also identified the subsidiary goals of fielding a first digitized division by 2000 and a first digitized corps by 2004, with the rest of the Army to follow.

The digitization of the Army has proceeded more slowly than was originally anticipated. As recently as 1999, the Army was planning to have 47 digitized brigades (32 active and 15 reserve) in the field by 2009. However, when planning began in 1999 for the new Objective Force of the future (following the installment of General Eric Shinseki as Army Chief of Staff), the idea of carrying digitization beyond the first digitized corps was abandoned and supplanted by new goals for the "transformed" army, which incorporated the objectives of digitization as a subset. (The Army's transformation involves making forces deployable more quickly while maintaining or improving their lethality and survivability.)

Under the umbrella of the Objective Force, the number of brigades to be digitized by 2009 has been reduced to 13--seven maneuver brigades in the first digitized corps and six new Stryker brigades. (Stryker brigades are interim units whose organization and capabilities place them between traditional brigades and Objective Force units.) By the end of 2003, the Army plans to have five digitized brigades. Four will be components of the 4th Infantry Division and 1st Cavalry Division in the first digitized corps, and the fifth will be the first Stryker brigade.

Bandwidth Supply at Army Commands Today

To understand how the Army intends to manage the communications networks and equipment that it now has--and the bandwidth they provide--one must consider how the hardware being used and the information coursing through it are handled within the Army's tactical command structure. The model of that structure used by CBO in its analysis is the structure of the first digitized division (the 4th Infantry Division) and the first digitized corps (the Third Corps). By adopting the Army's recent digitization framework, or architecture, as its starting point, CBO's analysis follows the Army's current IT strategy fairly closely, noting anticipated future changes in architecture as appropriate. In addition, while battlefield situations involve both wired and wireless (sometimes referred to as radio frequency) communications, this study focuses more on wireless messaging, because of its dominance in mobile battlefield operations and its relative cost. (Box 2 gives some sense of the interplay between the two types of messaging on the battlefield.)
Box 2.
The Evolving Mix of Wired and Wireless Battlefield Communications

Traditionally, battlefield communications systems have used radios (wireless communications), and very high bandwidth messaging made possible through the use of fiber-optic cable (wired communications) has been nonexistent. Wired communications using copper wires have long existed only in or near tactical operations centers (TOCs), with the principal purpose of tying together its local phone system. With such a system, the Army fought Desert Storm, and it faced no significant restrictions on its operations because of shortfalls in bandwidth.

In that conflict, the pace of the throughputs that radios provided for the command-and-control system (including what were then considered high-capacity satellite communications) was satisfactory because nearly all messages represented analog voice traffic for which the equipment and procedures were designed. U.S. forces were able to communicate much more freely than their adversaries could, and U.S. commanders' decisionmaking cycle was much faster as well. By the time of the war in Afghanistan, special forces personnel were using military variants of (wireless) cell phones to transmit voice and simple numerical data, such as global positioning system coordinates, to Air Force and Navy bombers.

In some battlefield situations in which the Army requires very high levels of bandwidth, it is available. Specifically, in combat environments where distances between operations centers are short, the nodes (in this instance, a group of colocated nets) move infrequently or not at all, and the risks of direct enemy attack are negligible, large demands for bandwidth are easily handled. In those circumstances, Army doctrine calls for running high-bandwidth land lines in the form of bundles of copper cable or using pulse-coded modulated (PCM) lasers, or both, to communicate between nodes. Also, within the TOCs of digitized units, fiber-optic cable is routinely used to connect elements of the internal local area networks, or LANs.1

Both PCM lasers and LANs support megabit-per-second throughput. But they are restricted in their geographic extent and limited to fixed nodes. The transmission range for a LAN is usually several hundred meters.2 PCM lasers are limited to about 20 kilometers because of their susceptibility to beam drift caused by wind and other atmospheric conditions. LANs and land lines typically avoid that problem, but they can take several hours to deploy.

In addition to those communications techniques, the Army's operational headquarters in Korea, Saudi Arabia, and Europe were linked within the past decade to the Pentagon and other U.S.-based headquarters using fiber-optic-based international commercial networks. Often called wide-area networks (WANs), they have become highly redundant in much of the world during the past 10 years. From a practical standpoint, however, WANs are restricted to fixed nodes.

Such approaches (WANs, LANs, PCM lasers, or bundled land lines) are generally of high bandwidth, but they are not practical for mobile vehicles and are therefore inappropriate for maneuvering forces that operate over wide areas. They are also less useful in cases in which a given message has a large intended audience or the precise locations of the recipients are unknown. In such instances, traditional (wireless) radios are ideal. They operate in broadcast mode, sending their signals out widely, and are useful for all levels of forces. Consequently, wireless, in spite of its lower intrinsic bandwidth relative to wired communications, and its much higher cost (25 to 50 times more) per unit of throughput, is the most practical communications medium for troops.

To supply its deployable forces, the Army has fielded hundreds of thousands of radios. The number of nodes that must in principle be linked by wireless technology to support an army on the move is about 100,000. To provide largely voice communications, the Army's inventory of radios in the field numbers about 220,000, or enough to support two major theater wars simultaneously.3

Wireless technology can reach the megabits-per-second range, and the Army currently has several programs designed to bring that rate of bandwidth to the individual soldier. The problem is that such high-bandwidth radios, designed to operate both on their own and with existing radios, may cost about $127,000 apiece.4 The minimum investment required to replace the entire fielded inventory of radios with new models is thus about $28 billion. In addition, because those replacement radios are still in development, that estimate of their cost is subject to a number of risks (discussed later), some of which could increase the investment substantially.

1.  That kind of connection has been demonstrated within TOCs associated with the 4th Infantry Division. Although in principle, a variety of LAN architectures are possible, the Army's intra-TOC LANs are organized as 100 Mbps Ethernets. (An Ethernet is a kind of LAN that is now recognized as an industry standard.)
2.  LAN geography is always limited, but the details depend on the speed of light in the fiber, the minimum synchronized operating frequencies of the nodes and other equipment on the network, the network's topology, and other physical characteristics.
3.  The figure 220,000 is also approximately the number of fielded radios capable of digital communications. In addition, the Army fields tens of thousands of older, mostly voice-only radios. However, the Congressional Budget Office has generally ignored those radios in its analysis because they are not part of the battlefield Internet and their contribution to overall communications bandwidth is small and decreasing.
4.  That average cost was noted in the December 2002 Selected Acquisition Report for the Joint Tactical Radio System, Cluster 1. The average unit procurement cost including research and development spending exceeds $140,000.

In the digitized Army, radios are organized into communications networks. As a shorthand, the networks are often referred to, synonymously, as nets or nodes. Army nets can be differentiated by the combat mission (operations, intelligence, logistics, fire support, and so on) that they predominantly serve. They are also characterized by the types of radios that produce their transmissions and by the other nets with which they directly communicate.

For instance, the operations officer in a battalion will manage communications from the operations net, usually designated the "ops" net. Today, the radios that this officer typically uses include an NTDR (Near-Term Data Radio), an EPLRS (Enhanced Position Location Reporting System), and one or more SINCGARSs (Single-Channel Ground and Airborne Radio Systems). At the battalion level, the operations officer (or a subordinate) directs large volumes of information to operations officers in the companies below and in the brigade above; takes orders and assessments from the brigade; and receives and transmits situation assessments from other operations officers in adjacent battalions.

Lesser volumes of information also pass to the ops net from the intelligence, fire-support, and other networks.(2) When such nets are colocated in the same command center, communications between them pass over a wired network, usually a local area network (LAN).(3) When nets are not colocated, which is common for communications between different levels of command or between forward and main headquarters, information from the intelligence, fire-support, and other nets that must be shared places additional demands on the ops net's wireless bandwidth. (Table 1 presents the equipment--mostly radios--that is now found in the operations nets of digitized Army units. The new and improved radios and other equipment that the Army plans to develop and field by the end of the decade are discussed in Chapter 2.)
Table 1.
Maximum Engineering and Effective Bandwidth for Typical Army Communications Equipment in 2003

(In kilobits per second)
  Point-to-Point Data Throughputs
Typical Battlefield
Command Levels

SINCGARS (SIP) Vehicle to Corps 16   1.7  
EPLRS (VHSIC) Company to Corps 128   13.3  
NTDR Company to Corps 288   30  
Interface Standardc Battalion to Army 16   1.5 to 7  
MSE Battalion to Corps 64   1.7  
MSE with ATM Switch Brigade to Corps 2,048d   5.1 to 6.7  
DSCS-111/93 Division to Army 256   27e  
DSCS-111/85 Division to Army 768   82e  
SMART-T Brigade to Army 4,620   481e  
STAR-Tf Corps to NCAg 24,000   2,500e  

Source: Congressional Budget Office based on the Army's 1999 budget hearing for its command, control, communications, and computer (C4) systems.
a. See the glossary of abbreviations.
b. These averages are lower than the maximum engineering throughputs because of the bandwidth required for context bits and channelization. The averages apply until about 2007, when the Army will begin to field the initial examples of a new generation of communications equipment (see Chapter 2 and Appendix A for more details). After a transition period between 2007 and 2010, Objective Force units in 2010 are scheduled to be the first units to incorporate the new equipment in its entirety.
c. Used for multiplexers, modems, routers, switches, radio access units, and other equipment.
d. The maximum potential rate is 8.192 megabits per second using the (high-capacity line-of-sight) HCLOS radio--provided the interfaces are programmed to operate at the higher rates and frequencies are available. However, at present, the interfaces are not so programmed.
e. Extrapolated from the reductions in bandwidth that occur for lower-frequency radios.
f. The termination in 2001 (for default, as a result of delays and cost overruns) of the ongoing contract for the STAR-T has cast doubt on the program's future. To fill the void produced by the termination, the Army is currently using commercially available systems of approximately the required throughputs and considering replacement candidates.
g. The NCA, or National Command Authority, refers to the command chain that extends to the Secretary of Defense and the President.

The effective communications bandwidth provided by the Army's voice and data radios is typically reduced below the maximum (engineering) rates by a factor of about 10, owing to the bandwidth needed for context bits and channelization (see Box 3). Because the mobile subscriber exchange, or MSE (the equivalent of a private telephone system), has a much higher degree of channelization than any of the other radios, its throughput is reduced by roughly a factor of 100 for point-to-point transmissions.
Box 3.
Operational Versus Theoretical Bandwidth

Operational point-to-point throughputs are always significantly lower than their theoretical maximums. That reduction stems from two factors: context bits and channelization.

Context Bits
For military radios, the requirement for context bits for digital throughput is an amalgam of a number of elements. Most of them are simply the result of trade-offs among the engineering constraints associated with hardware, software, and waveforms. Conceptually, however, most of the reduction in throughput attributable to context bits derives from the following requirements and considerations:
  • Network Management. The Army's tactical Internet uses standard commercial Internet formats such as http, TCP, and IP.1 Each format has a somewhat different overhead, data-packing format, and packet-transmission protocol.

  • Network Architecture. Particularly at higher headquarters levels, a network's performance in terms of throughput depends on the network's topology--the number of nodes, the router configurations, the numbers and linkages of relays, and the degree to which messages are multicast (sent to multiple nodes simultaneously). In practice, the network's performance degrades as the number of internal nodes, relays, and routers increases.

  • Forward Error Correction and Retransmission. Overhead (for example, parity bits), checksum generation, and the time spent by the packet transmitter and receiver in checking errors cut bandwidth even when error rates are low. When error rates are high, message retransmissions also reduce effective bandwidth. Forward error correction, which in its simplest form involves parity bits and some form of data redundancy, allows many errors to be fixed without initiating retransmissions. However, when forward error correction fails, packet retransmissions may be initiated by both the transmitter and the receiver. Although often transparent to the communications system's user, packet retransmission requests and packet retransmissions often dominate message traffic at longer ranges where "noise" becomes a significant problem.

  • Encryption and Decryption. Throughput can be slowed if encryption/decryption devices do not operate fast enough to deal with the message traffic.

  • Frequency-Hopping Overhead. Radios can be designed to randomly change the frequencies at which they are transmitting to make it harder to detect and jam transmitted signals. The processing associated with frequency hopping reduces effective throughput.

  • Time-Hopping Overhead. Radios can be designed to randomly stagger their transmissions to make signal detection and jamming more difficult. The processing associated with those delays reduces effective throughput.

  • Software Inefficiencies. Hardware devices and the software that makes them function are designed according to certain assumptions about the communications network in which they will operate and the rates of information flow that will apply. As networks age (that is, become loaded with more and newer devices), those assumptions about the network configuration may become invalid, which can lead to reduced throughput.

    Channelization is the term used to express the number of different data flows a radio can simultaneously support. Simpler radios like the Single-Channel Ground and Airborne Radio System will support either one data channel or one voice channel but not both at once. (The operator flips a switch to use one or the other.) The mobile subscriber exchange--the Army's digital phone system--can be manually set to simultaneously handle large numbers of phone calls (high channelization) or reconfigured to provide one higher-bandwidth channel (low channelization). Low channelization might be used to support a video teleconferencing transmission. The greater the channelization, the lower the effective point-to-point throughput will be.

    1.  Respectively, hypertext transfer protocol, transmission control protocol, and Internet protocol. Each format is optimized differently for different types of data.

The operations officer at a given level of command has one or more radios, or "pipes," available for communications (see Table 2). The approximate capacity of what is termed a trunk line at a particular command level (essentially, the operations network) can be obtained by adding up the capacities provided by the available individual pipes.(4)
Table 2.
Number and Types of Radios Available at the Ops Nodes in Digitized Units During Peak Operations in 2003, by Command Level


Corps 1 1 1 1
Division 1 1 1 1
Brigade 1 1 1 1
Battalion 1 1 1 0
Company 1 1 0 0
Platoon 1 1 0 0
Squad/Vehicle 1 0 0 0

Source: Congressional Budget Office based on Department of the Army, Army Tactical Communications (1999), which details the architectural design of the digitized Army.
Notes: In addition to the radios, or "pipes," noted above, which form the trunk lines for the flow of digital information, operations officers may sometimes use analog voice-only systems such as walkie-talkies or several types of purely analog radios. Those analog systems carry a communications load that is not significant for CBO's analysis.
The numbers represent the average number of pipes allocated to the operations officer. For instance, in the column labeled "Satellite Links," a digitized brigade command center usually has three independent satellite terminals (and satellites) available. However, during peak operational conditions, intelligence officers and fire-support activities are typically using two of the terminals--leaving only one, on average, for the operations officer.
See the glossary for abbreviations.
a. At the higher command levels, the table refers to the operations networks only. At lower levels, the distinctions between the various communications networks (for example, operations, intelligence, and fire-support) become less clear.

Combining the information in Tables 1 and 2, CBO estimated the maximum total effective throughput that the Army's ops nets can provide. To do that, CBO made several assumptions (which represent ideal circumstances): for a given point-to-point communication, the throughput load can be fitted to perfectly match the maximum operational capabilities of the available radios; input and output throughputs are equal;(5) and all intended recipients are within an average operational range of all the radios. Under such assumptions, the data throughput rates of individual radios can be added to estimate total throughput capability--the total bandwidth supply--provided by the radios available in each ops net (see Table 3).
Table 3.
Maximum Effective Bandwidth Available to Army Operations Networks in 2003, by Command Level

(In kilobits per second)
Command Levela Total Effective Throughputb

Corps 2,550  
Division 533  
Brigadec 533
Battalion 37  
Company 15  
Platoon 15  
Squad/Vehicle 1.7  

Source: Congressional Budget Office.
a. At the higher command levels, the table refers to the operations networks only. At lower levels, the distinctions between the various communications networks (for example, operations, intelligence, and fire-support) become less clear.
b. Point to point, under an assumption of perfect load balancing. In practice, throughput rates may be less.
c. The up-arrow (↑ ) indicates the throughput rate for communications to equivalent or higher command levels. The down-arrow (↓ ) indicates the throughput rate to lower command levels.

As the table shows, the digitized brigade has two measures of available bandwidth. That command level employs both the Army's beyond-line-of-sight (BLOS) satellite communications and its line-of-sight (LOS) radio communications. The bandwidth provided by BLOS satellite communications is several orders of magnitude greater than that available to battlefield LOS radios. The brigade can communicate with higher commands using the BLOS transmitters, as signified by the up arrow. However, it must talk to lower commands with LOS NTDRs, indicated by the down-arrow, because digitized battalions do not have satellite receivers. That satellite communications distinction also divides the upper tactical Internet (which is available to commands at the brigade level and above) from the lower tactical Internet (available to the brigade and lower command levels).

Bandwidth Demand at Army Commands Today

Experiments have shown that message traffic and data that provide situation awareness and support for decisionmaking, including digital telephone calls, are the bulk of the information transmitted in the Army's communications system. Communications traffic can be thought of as either approximately continuous or approximately episodic. In the former case, called continuous-flow information, bits per second is the relevant measure; in the latter case, referred to here as episodic information, the size of the message file (in bits) is the appropriate gauge. For instance, situation assessments of U.S. forces that are arriving asynchronously every minute or so from a number of points are best represented as a continuous flow of information. By contrast, a command for a dynamic retasking (also known as a dynamic unit task order, or UTO) may occur only once a day.(6)

Bandwidth Demand for Continuous-Flow Information

The Army has studied the impact of digitization on peak continuous bandwidth demand at the division, brigade, and battalion levels. (For details of that study, see Appendix B.) For continuous-flow information, the digitized division generates a peak demand on the operations net of between 2.5 Mbps and 4 Mbps. Major consumers of bandwidth are ordinary telephone traffic, the Army Battle Command System (ABCS), and video teleconferencing (see Table 4).(7) The ABCS is differentiated into segments for classified and unclassified information.
Table 4.
Peak Demand for Continuous-Flow Bandwidth in the Digitized Division's Operations Net in 2003

(In kilobits per second)
Source of Demand Size of Deman

Army Battle Command System  
  Classified, nonlogistics traffic 300 to 1,000  
  Unclassified, mostly logistics traffic 100 to 300  
Telephone (Digital)a 1,400  
Unmanned Aerial Vehicles 100 to 300  
Video Teleconferencing 1,000  

Source: Congressional Budget Office based on data from the Department of the Army (see Appendix B for details).
a. Analog telephone communications employing a number of analog radio and walkie-talkie systems are also used but are not considered in this analysis because they are either independent of the communications trunk lines or contribute little to throughput demand.

Although data on bandwidth demand are available for the brigade and battalion levels of command, similar data do not exist for the other command levels. Nonetheless, peak continuous demand for them can be extrapolated from the data available for the three higher command levels. The key assumption in the extrapolation is that information requirements change from command level to command level in proportion to alterations in the span of control over resources and personnel. Since resources and personnel increase by roughly a factor of three at each command level, so, too, will information requirements. In its estimates of bandwidth demand for continuous-flow data, therefore, CBO assumed that at the corps level, the demand for bandwidth was three times the demand at the division level and that each command below the battalion level required one-third less bandwidth than the next higher command.

That "factor-of-three" assumption has been analyzed both in this study and by the Army (see Appendix B for details). Data on information flows at several different levels of command, although incomplete, were analyzed in one of the Army studies surveyed for this report; the results suggest that the actual scaling factor probably lies between two and four.(8) However, for the sake of specificity and because the results of the analysis remain qualitatively the same, CBO used a factor of three in its analysis. (Table 5 presents the results of applying the factor-of-three extrapolation to the data in Table 4.)
Table 5.
Peak Demand for Continuous-Flow Bandwidth in 2003, by Command Level

(In kilobits per second)
Army Battle Command System Updates
VTCb UAV and
Classified Unclassified

Corps 1,000 to 3,000   300 to 1,000   4,000   1,000   100 to 300  
Division 300 to 1,000   100 to 300   1,400   1,000   100 to 300  
Brigade 100 to 300   30 to 100   300   300   100 to 300  
Battalion 30 to 100   n.a.   100   300c   100 to 300  
Company 10 to 30   n.a.   30   n.a.   n.a.  
Platoon 3 to 10   n.a.   10   n.a.   n.a.  
Squad/Vehicle 1 to 3   n.a.   3   n.a.   n.a.  

Source: Congressional Budget Office based on data from the Department of the Army (see Appendix B for details).
Notes: The estimates of bandwidth demand in this table were calculated by applying a "factor-of-three" extrapolation to the data in Table 4.
UAV = unmanned aerial vehicle; n.a. = not applicable for that command level.
a. At the higher command levels, the table refers to the operations networks only. At lower levels, the distinctions between the various communications networks (for example, operations, intelligence, and fire-support) become less clear.
b. Video teleconferencing and UAVs have bandwidth requirements that are independent of command level.
c. In the digitized battalion, collaborative planning, which is a component of the classified Army Battle Command System, and video teleconferencing are not in operation simultaneously.

Bandwidth Demand for Episodic Data

Episodic transmissions on the battlefield fall into several categories: fragmentary orders (FRAGOs), operations orders (OPORDs), fused intelligence reports, UTOs, and map resynchronizations.(9) As noted earlier, the relevant measure of such messages is the size of the file being sent, which CBO estimated on the basis of data gleaned from the advanced warfighting experiments (AWEs) held between 1998 and 2001 at the brigade and division levels and from subject matter experts in communications. Collectively, those experts, who were drawn mainly from the operations, C4 (command, control, communications, and computers), and resources offices of Army headquarters, have experience at multiple command levels generating and using such types of episodic messages (see Table 6).
Table 6.
File Sizes of Episodic Throughputs in Army Ops Channels in 2003, by Type of Throughput and Command Level

(In kilobits)

Corps 100   10,000   100   1,000   3,000 to 10,000  
Division 100   1,000   10 to 100   1,000   1,000 to 3,000  
Brigade 10   100   10 to 100   1,000   300 to 1,000  
Battalion 1   10   10 to 100   1,000   100 to 300  
Company 1   10   10 to 100   100   30 to 100  
Platoon 1   10   10 to 100   100   10 to 30  
Squad/Vehicle 1   10   10 to 100   100   3 to 10  

Source: Congressional Budget Office based on data from Army subject matter experts and from the advanced warfighting experiments held between 1998 and 2001 at the brigade and division levels.
a. At the higher command levels, the table refers to the operations networks only. At lower levels, the distinctions between the various communications networks (for example, operations, intelligence, and fire-support) become less clear.
b. Usually short amendments to operational orders, which are transmitted as character data.
c. Issued once or twice daily, operational orders comprise character-based information including force subordination, disposition of the enemy, the commander's intent, operational goals for the commander's force elements, and decision branch points.
d. Map resynchronizations, or "resyncs," occur because information associated with digital maps must be updated when a unit moves from one map sector to another.

Calculating peak demand for continuous-flow data at all command levels requires an estimate of the average frequency with which episodic throughputs occur. But no such data have been collected. Therefore, CBO has again generated assumptions, based on information from Army subject matter experts, about such frequencies. FRAGOs and fused intelligence reports are assumed to arrive once a minute at commands below the brigade level and once an hour at commands above it. OPORDS, UTOs, and map resychronizations are assumed not to occur during peak operations.

Using those assumptions, CBO calculated the equivalent peak continuous information flow associated with those episodic transmissions. Bandwidth demand for those messages, CBO estimates, would be significant only at the lowest levels of command (see Table 7). CBO then combined the estimates for both continuous-flow and episodic transmissions to yield an estimate of total continuous bandwidth demand at each command level (see Table 8).
Table 7.
Equivalent Peak Continuous-Flow Bandwidth for Episodic Throughputs in 2003, by Throughput Type and Command Level

(In kilobits per second)
Unit Task

Corps 0.03   0   0.03   0   0  
Division 0.03   0   0.003 to 0.03   0   0  
Brigade 0.003   0   0.003 to 0.03   0   0  
Battalion 0.017   0   0.17 to 1.7   0   0  
Company 0.017   0   0.17 to 1.7   0   0  
Platoon 0.017   0   0.17 to 1.7   0   0  
Squad/Vehicle 0.017   0   0.17 to 1.7   0   0  

Source: Congressional Budget Office based on data from Army subject matter experts and from the advanced warfighting experiments held between 1998 and 2001 at the brigade and division levels.
a. At the higher command levels, the table refers to the operations networks only. At lower levels, the distinctions between the various communications networks (for example, operations, intelligence, and fire-support) become less clear.
b. Usually short amendments to operational orders, which are transmitted as character data.
c. Issued once or twice daily, operational orders comprise character-based information including force subordination, disposition of the enemy, the commander's intent, operational goals for the commander's force elements, and decision branch points.
d. Map resynchronizations, or "resyncs," occur because information associated with digital maps must be updated when a unit moves from one map sector to another.

Table 8.
Total Peak Demand for Effective Bandwidth in 2003, by Command Level

(In kilobits per second)
Command Levela Peak Bandwidth Demand

Corps 3,000 to 10,000  
Division 2,500 to 4,000  
Brigade 800 to 1,300  
Battalion 500 to 750  
Company 30 to 100  
Platoon 10 to 30  
Squad/Vehicle 3 to 10  

Source: Congressional Budget Office.
Note: Demand is extrapolated from the division level.
a. At the higher command levels, the table refers to the operations networks only. At lower levels, the distinctions between the various communications networks (for example, operations, intelligence, and fire-support) become less clear.


Comparing Bandwidth Supply and Demand in 2003

CBO compared the maximum effective bandwidth available at the operations desks of various tactical command levels (shown in Table 3) with its estimates of total demand at those levels (from Table 8). The results of CBO's comparison show that at no command level is there currently a substantial excess of supply relative to demand. In the worst case (for communications from the brigade to lower command levels), demand exceeds supply by an order of magnitude or more (that is, by more than 10 to one). In the best case (at the platoon level), demand is somewhere between one-half and twice the bandwidth supply.

To strengthen the qualitative aspects of CBO's comparisons, color coding has been used in Table 9. At one extreme--for communications from the brigade ops net to operations desks at lower levels and for the battalion ops net--red implies that demand exceeds supply by a factor of 10 or more. A cautionary yellow reflects a match between supply and demand to within a factor of three. (Caution is warranted because although, on average, messages can be expected to get through the network on time, some delays in transmission may be experienced.) Shades of orange indicate gradually worsening imbalances between supply and demand, lying between the cautionary yellow range and the substantial mismatch indicated by red. (In the event of results showing that supply exceeds demand by at least a factor of three, green will be used.)
Table 9.
Effective Bandwidth Supply Versus Peak Demand in 2003, by Command Level

(In kilobits per second)
Relative Supply
Versus Peak
(S : D)b

Corps 2,550   3,000 to 10,000   1 : 1 to 4  
Division 533   2,500 to 4,000   1 : 5 to 8  
Brigadec 533 800 to 1,300   1 : 1.5 to 3
  37     1 : 20 to 30
Battalion 37   500 to 750   1 : 10 to 20  
Company 15   30 to 100   1 : 2 to 6  
Platoon 15   10 to 30   1 : 0.5 to 2  
Squad/Vehicle 1.7   3 to 10   1 : 2 to 6  

Source: Congressional Budget Office.
a. At the higher command levels, the table refers to the operations networks only. At lower levels, the distinctions between the various communications networks (for example, operations, intelligence, and fire-support) become less clear.
b. Based on an approximate logarithmic scale, the color coding is as follows: yellow indicates that supply is between about one-third and three times demand (a marginal supply/demand match); light orange signifies that demand is approximately three times supply and orange, that demand is approximately three to 10 times supply. Red (used here for the lower brigade-level relationship and at the battalion level) means that demand exceeds supply by a factor of 10 or more.
c. The up-arrow (↑ ) indicates the throughput rate for communications to equivalent or higher command levels. The down-arrow (↓ ) indicates the throughput rate to lower command levels.

How does the Army's experience compare with CBO's results? The Army has carried out relatively little testing of large operations nets. However, when it has, participants have invariably cited shortfalls in bandwidth supply as a significant problem. The Army's advanced warfighting experiments conducted at the National Training Center in 1997 and 1998 were, respectively, battalion- and brigade-level experiments using state-of-the-art communications equipment. The AWEs revealed bandwidth problems and network failures to the point where soldiers switched back to analog voice communications as the transmission of digital data slowed. During the Division Capstone Exercise, which was undertaken in April 2001 at the culmination of the four-year effort to develop the digitized division, computer crashes occurred that were attributed to overloaded communications systems.

Some insight is also available from more limited tests of networks in the development stage. For instance, at Ft. Hauchuca in 1999, a company-level test evaluated a limited communications network equipped with the Army's most advanced communications gear. Severe degradation occurred in the rate of message completion--it dropped to less than 60 percent--as the network's message load was increased to peak operational levels. In particular, for Blue situation assessments that were successfully transmitted, the latency, or delay, averaged four minutes--a rate that may compromise certain missions.

Long latencies--delays in message transmission or receipt--are often observed in the context of exercises. However, shortages of bandwidth may not only produce latencies in communications networks but in some cases also exacerbate other causes of delays. Operations officers remark that although long latencies are not a handicap for deliberate planning or low-intensity operations, they preclude reliance on the network "when the shooting starts."

Army studies have attempted to estimate the performance of operations networks for large units in the field. One of the studies hints at the existence of the bandwidth bottleneck that CBO's analysis projects for communications from the brigade to the battalion:

"Our analysis revealed that when large secondary imagery dissemination products are provided down to battalion level, the maximum data rate capability of the radio [NTDR] is utilized. This leaves no reserve capacity to the battalion and confirms the need for more data bandwidth between the brigades and battalions."(10)

Operation Iraqi Freedom

The Army's plans for the recent conflict in Iraq included the use of at least one digitized unit in the early phase. The 4th Infantry Division, a unit of the 5th Corps, employs the communications architecture described in this study. The division was originally slated to deploy through Turkey and then be involved in the early portions of the conflict. Instead, when it finally disembarked in Kuwait, the conflict was deemed to have largely ended. Therefore, even a qualitative comparison of the results of CBO's analysis and the division's experience with bandwidth supply and demand in combat is impossible.

Yet reports indicate that bandwidth was an issue for those units that were engaged in the conflict. In the months prior to the war and in anticipation of operating in combat with the 4th Infantry Division, other units in the 5th Corps, selected units in the Marine Corps's 1st Marine Expeditionary Force, and the British armored division were outfitted with 1,000 equipment sets for interfacing with the 4th Infantry Division. Contractor support was also provided to enable and sustain communications. The resulting system is now called Blue Force Tracker.

The new system did not fully "digitize" any of those other units, however. Instead, their degree of digitization was termed "digitization lite" by Army officers assigned to quickly provide the interfacing equipment. Typical sets, some of which went to forces at the company level, comprised one terminal from the Force XXI Battle Command, Brigade and Below (FBCB2) program, a Global Command and Control System (GCCS) terminal at the higher command levels, an ABCS terminal, and a commercial L-band satellite transceiver (now increasingly common among U.S. trucking firms), together with interfacing gear.(11) By using commercial L-band satellites (the military has not invested heavily in the L-band), widely dispersed commands could keep track of forces that had the GCCS and FBCB2 displays; at the higher command levels, portions of the GCCS supported planning and decisionmaking. (However, the Army could not fully utilize the ABCS because of unmet National Security Agency certification requirements and other inadequacies.)

When not challenged by line-of-sight constraints, soldiers used their SINCGARS and EPLRS equipment for voice communications. When the widely dispersed and rapidly moving forces faced LOS problems, soldiers substituted commercial e-mail and "chat room" messages for point-to-point and collective voice communications. The mobile subscriber exchange proved useful between fixed communications sites but much of the time was not mobile enough for troops on the move.

At the tactical operations centers at the higher command levels, the minimal equipment configuration described above supplied digital bandwidth that was about one-quarter to one-third of that available in fully digitized units. On the demand side, there was no basis for comparison, for several reasons: during the conflict, trunk lines were often "saturated"--all available digital bandwidth was used up; demand was not subject to the monitoring afforded in tests; lower command levels were not equipped with the FBCB2; and the full capabilities of the ATCCS could not be exploited.

Despite those drawbacks, some insights might be gained. Fully coordinated Department of Defense "official lessons" from Operation Iraqi Freedom are not yet available, and quantitative data may never be. But qualitative statements from some authoritative sources have been issued.(12) Limited bandwidth was a constant problem in spite of the large increase relative to that available during Desert Storm; bandwidth from commercial satellites in particular was much more heavily used.(13) Although the Army has made a substantial investment in military-only decision-support systems, much of the planning and collective decisionmaking that occurred during the Iraq war was handled through commercial e-mail and chat-room applications that soldiers were familiar with, that were "user friendly" and reliable over long distances, and that required little or no training. Another factor driving forces to use chat-rooms and e-mail was that the distances over which messages had to be transmitted precluded the use of LOS radios and rather than wait or hope for reception to improve (as a result of more relays being deployed), they turned to satellite communications to speed their operations.

Blue Force Tracker was praised by most users in both the Army and the Marine Corps.(14) However, the rapid pace of much of the operation and the wide dispersal of U.S. forces made LOS communications challenging, which, as noted earlier, rely on equally rapid and widely dispersed relays to remain interconnected. (As one example, shortfalls of bandwidth for such communications were nevertheless noted as "information overload," despite the limited information flow those applications--mainly e-mail and chat rooms--generated and the relatively small number of network users.)(15) The formats provided by Blue Force Tracker ranged from bandwidth-intensive imagery to low-bandwidth e-mail text messages. But because bandwidth saturation affected communications for both the Army and Marine Corps, users were often forced to employ message formats that used the least possible bandwidth (in other words, text messages).

Other observations during the operation were related to the availability and survivability of the bandwidth supply. Battery deficiencies (electrical generators are too slow to deploy with rapidly moving forces) were noted, particularly at executive command levels, and concerns were raised about the communications systems' electronic "footprint."(16)

1.  Monica Farah Stapleton and Yosry Barsoum, "C4ISR Systems Engineering Analyses and Modeling & Simulation" (briefing prepared for the Army Communications and Electronics Research, Development, and Engineering Center by AMSAA [Army Materiel Systems Analysis Activity] and the Mitre Corporation, September 9, 2002); J.L. Burbank and others, "Concepts for the Employment of Satellite Communications in the Army Objective Force" (draft, Johns Hopkins University Applied Physics Laboratory, August 2002); RAND Arroyo Center, "Future Army Bandwidth Needs--Interim Assessment" (briefing prepared for the G6/Army Chief Information Officer, July 10, 2002); Yosry Barsoum, "Bandwidth Analysis (ACUS Only) of Division Main, Maneuver Brigade TOC, and Tank Battalion" (briefing materials prepared for the Army's Communications and Electronics Command by the Mitre Corporation, February 29, 2000); Steve Chizmar and others, "Digitization at Brigade and Below (DB2) Study" (briefing prepared for the Army Materiel Systems Analysis Activity, Aberdeen Proving Ground, Maryland, December 1, 1999); Army Medical Department Center and School, "Medical Force Digitization Overview" (briefing prepared for the Deputy Chief of Staff for Command, Control, Communications, and Computers, January 1999). The following were provided to CBO by the staff group of the Army Chief of Staff: Army Signal Center and Fort Gordon, Directorate of Combat Developments, Modeling and Simulation Branch, Architecture Division, First Digitized Forces System Architecture (1DFSA): Version 2.02, Simulation Analysis/Study, White Paper (main text and attachment titled "Satellite Communications Capacity Study," June 30, 1999; released to CBO on May 14, 2002); and Lt. Gen. P. Cuviello, "Projected Bandwidth Usage and Capacity" (briefing prepared for the Army Chief of Staff by the G6/Army Chief Information Officer, August 2002; released to CBO on January 14, 2003).
2.  Those networks carry information internal to their mission areas and analogous in size to the large volumes passing through ops channels.
3.  A LAN is one example of a network that is "wired," or connected, by very high bandwidth fiber-optic cable.
4.  If the demand for information through those pipes could be exactly matched to their size, then the trunk line's capacity would exactly equal the sum of the pipes' bandwidths. In practice, however, that does not always occur; therefore, the sum represents an upper bound on the trunk line's actual bandwidth. The trunk line for digital communications does not include a number of older, generally analog radios that carry relatively little message traffic.
5.  For the most part, that assumption is valid for ops nets, which according to Army doctrine share information with both higher and lower command levels. For other nets, the assumption may not apply. Fire-support nets, for example, often require much higher bandwidth for inputs than for outputs because their incoming traffic is dominated by data from high-bandwidth sensors such as UAVs, whereas their outgoing traffic is made up of relatively low bandwidth messages telling troops to fire at specific targets.
6.  A requirement for a dynamic retasking occurs when a unit is "chopped" from one commander to another--for instance, when a battalion that is subordinate to one brigade at the start of battle is reassigned to another brigade during the course of an operation. Many command and logistics responsibilities are severed and reestablished in such a retasking, requiring the reconfiguration of many network files (for example, address lists and control files).
7.  The ABCS is a so-called system-of-systems created from the Army's Tactical Command and Control System (ATCCS) and the Force XXI Battle Command, Brigade and Below (FBCB2) program. The ATCCS is one of the original command-and-control system-of-systems. Created in 1979, it serves as an organizational umbrella for several Army command-and-control programs such as the All Source Analysis System and the Army Field Artillery Tactical Data System, which were initiated in 1969 to manage intelligence and artillery data, respectively. FBCB2 is a system of hardware, software, and databases providing automated capability for situation awareness and command and control to brigade and lower command levels.
8.  Barsoum, "Bandwidth Analysis (ACUS Only)." The data and analysis are discussed in Appendix B.
9.  FRAGOs are usually short amendments to operational orders and are transmitted as character data. OPORDs are rather lengthy operational orders that are issued once or twice daily. They comprise character-based information including force subordination, disposition of the enemy, the commander's intent, operational goals for the commander's force elements, and decision branch points. Fused intelligence reports are the intelligence summaries produced by intelligence officers that are shared with operations officers. Map resynchronizations, or "resyncs," occur because information associated with digital maps must be updated when a unit moves from one map sector to another.
10.  Army Signal Center and Fort Gordon, Directorate of Combat Developments, First Digitized Forces System Architecture, p. 1.
11.  The GCCS comprises both hardware and applications-level software on the network.
12.  The following discussion draws heavily on the after-action reports (AARs) issued by major units engaged in the conflict: those of the 5th Corps (entitled "V Corps: C4ISR Integration AAR"), the 3rd Infantry Division (see Chapters 17 and 26), and the 2nd Brigade, 101st Airborne Division.
13.  At the peak of the conflict, the Defense Information Systems Agency claimed that 3 Gbps of satellite bandwidth was being provided to the theater, 84 percent of which was commercial. That amount is 30 times the satellite bandwidth made available during Desert Storm. See "DISA Chief Outlines Wartime Successes," Federal Computer Week, June 6, 2003.
14.  One senior Marine Corps officer told CBO that while arguments might continue over the lessons learned from Operation Iraqi Freedom, there seemed to be community-wide agreement on Blue Force Tracker. To paraphrase, "We used to lose units all the time. Subject to bandwidth constraints, with Blue Force Tracker we could find them almost immediately. . . . We're going to buy more."
15.  J. Davis, a reporter embedded with the Army's 11th Signal Brigade in southern Iraq, reported in the June 2003 issue of Wired Magazine that "[b]ecause anyone on Siprnet [a DoD classified network] who wanted to could set up a chat, 50 [chat] rooms sprang up. . . . The result: information overload." In the same article, Lt. Col. N. Mims of the 11th Brigade is quoted as saying, "We've started throwing people out of the rooms who don't belong there."
16.  An electronic footprint is the size, or power, of electromagnetic emissions at various distances. J. Burias, in "G-6 Says OIF [Operation Iraqi Freedom] Validates IT Transformation Path," Army Link News, May 30, 2003, quoted Lt. Gen. P. Cuviello: "Antenna farms sprang up around major Army units in both Afghanistan and Iraq as different antennas were needed for each of six different satellite bands and four different types of radios. . . . All those antennas sometimes caused co-site interference with each other." Mobility requirements in Iraq and Afghanistan forced troops to rely heavily on batteries rather than the generators normally used at fixed locations. In the same article, Cuviello is also quoted as saying, "Batteries are heavy items to carry around the battlefield--not only to keep them stocked and transported, but also the transportation requirements to dispose of them."

Previous Page Table of Contents Next Page