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

Mitigating Mismatches Between
Bandwidth Supply and Demand

The Congressional Budget Office has examined several ways to treat the projected mismatches between the Army's demand for and supply of communications bandwidth on the battlefield. Those approaches include increasing supply, decreasing demand, reallocating currently planned spending to focus more resources on command levels with the most severe bandwidth problems, and developing tools to better manage whatever mismatches persist. Increasing the supply of bandwidth by buying more or better radios would probably be the least effective way to mitigate the supply/demand imbalance. Indicators of the efficacy of reallocating resources are ambiguous, and their results depend on the metric being used. CBO is thus unable to provide any unequivocal findings on that option. Decreasing demand would help lessen the mismatch, and some potential options are discussed, although they would not eliminate the imbalance. Better management of supply and demand would also help shrink the supply/demand gap, and several aspects of that approach are noted.

Buy Better Radios in Greater Quantities

The Army might increase its supply of bandwidth in 2010 by buying radios that are more technologically advanced than those it is currently expecting to have. Yet the service's planned programs appear to already be incorporating all of the most likely advances in communications technology. To develop new radios such as the Joint Tactical Radio System, the Army has contracted with large corporate teams that include many of the world's premier radio makers and computer network experts, and program managers report that those development efforts are fully funded.(1) Because of the ambitiousness of some of the technological advances being pursued, there is a risk that not all of them will be realized (see Appendix A). Thus, it seems unlikely that advances in technology over and above those that the Army is already pursuing could reduce the bandwidth supply/demand mismatches that CBO projects.

Also questionable is whether the projected shortfalls in the supply of Army bandwidth could be eliminated by purchasing more of the communications equipment that is now being developed. At the division and corps levels, CBO estimates, demand in 2010 will outstrip supply by a factor ranging between 10 and 30. At those levels of command, the Warfighter Information Network-Tactical program will supply the majority of bandwidth, and the Army has considered increasing its purchases of WIN-T equipment 20-fold to better match the expected demand. But implementing such an option would present a number of challenges, including management of the increased complexity and inefficiency of the communications network that would result; the physical space, or "footprint," required to accommodate the additional equipment on the battlefield; and the additional cost.

Increased Network Complexity and Costs

As the number of nodes in a communications network increases, the network's efficiency--measured as the throughput capacity available at each node--decreases (see Figure 1). As the figure shows, an order-of-magnitude increase in the number of nodes contained in the network can produce nearly an order-of-magnitude decrease in bandwidth at a given node. The results in the figure are consistent with theoretical estimates of how a network's efficiency (measured in kilobits per second) varies with the number of nodes when current methods (specifically, Internet processing protocols) for transmitting information around a network are used.
Figure 1.
Notional Throughput Capacity per Node

(In kilobits per second)
Source: Congressional Budget Office based on a briefing on the upper and lower tactical Internets provided by the Army Chief of Staff, Staff Group.
Note: The rate of throughput is highly dependent on the number of nodes in the network.

Theory indicates that network throughput can be expected to decay as 1/Q, where Q is the square root of Nln(N), N is the number of nodes in the network, and ln is the natural logarithm.(2) Consequently, if the number of radios increased by a factor of 20, the network's throughput would rise by only a factor of 20/7--or about 3--where 7 is the reduction in throughput caused by the increase in the network's complexity.(3) In other words, the 20-fold increase in purchases of WIN-T equipment that the Army has considered would not be sufficient to overcome the 10- to 30-fold shortfall in available bandwidth that CBO projects for 2010 at the corps and division levels of command. Theoretically, at least a 600-fold increase would be required, which the Army is unlikely to pursue, given that WIN-T procurement costs are currently estimated to be about $5 billion.(4)

Expansion of the Footprint

The geographic footprint of a program is the physical space that the program's equipment occupies when it is operated on the battlefield. The Army's current plans for the WIN-T and JTRS have raised concerns about the proliferation of equipment in vehicles and operations centers. Part of the reason for that concern is a lack of physical space, an element that has also surfaced in discussions about changing the network architecture. Another issue is that the associated increase in the equipment's electronic footprint (the size, or power, of its electromagnetic emissions at various distances), caused by the proliferation of radiating antennas and electronic "noise," will facilitate enemies' detection and targeting of U.S. forces. In that context, increasing the amount of communications equipment by a factor of 20--or more--could pose significant problems.

Reallocate Currently Planned Spending

If additional spending is unlikely to close the gap between bandwidth supply and demand, might currently planned spending be reallocated to focus more on the command levels that projections show will have the most severe problems in 2010? Answering that question requires constructing so-called figures of merit to relate projected spending at each command level to communications capacity--both the capacity that will be provided and the capacity that will be demanded. Several such metrics can be considered.

Spending data indicate that over the next several years, the Army plans to spend more money at the lower levels of command--that is, below the division and corps levels and away from the bandwidth bottleneck projected for 2010 (see Figure 2). That distribution of expenditures arises because radios are spread throughout the force and most personnel with radios are located not at higher command-level headquarters but in the decreasingly smaller, more numerous units that compose the lower levels of command.
Figure 2.
Total Projected Investment in Ops Net Equipment, by Command Level

(In billions of fiscal year 2003 dollars)
Source: Congressional Budget Office.

Considering the cost of the total bandwidth capability that the Army plans to provide yields a different picture. Lower levels of command comprise many vehicles and soldiers, which the Army envisions equipping with relatively simple versions of the JTRS. At higher command levels, smaller numbers of more complex and more expensive versions of the JTRS will be fielded, as well as more sophisticated--and costly--line-of-sight and beyond-line-of-sight satellite communications radios. That latter equipment will supply much greater amounts of bandwidth per unit and support more networks than the simpler but more numerous radios planned for the lower levels of command.

A metric can be constructed that accounts for the cost of the capacity provided per ops net at each command level (see Figure 3). As the figure shows, the cost is greatest for the levels of command at which the bandwidth bottleneck is projected to be most severe in 2010.
Figure 3.
Cost of the Capacity Provided per Ops Net, by Command Level

(In fiscal year 2003 dollars/bit per second/net)
Source: Congressional Budget Office.

Considering the cost of the capacity utilized (as opposed to provided) yields yet another view of how the Army's projected investments are distributed by command level (see Figure 4). As Chapter 2 discussed, the supply of bandwidth at the squad and platoon levels in 2010 will exceed the projected demand by factors ranging from four to 20. The reason is that the simpler version of the JTRS to be fielded at lower command levels is equipped to communicate with higher levels; that is, it has the capability to use the wide-band networking waveform (discussed in Appendix A), which provides substantial bandwidth--more than the demand now being projected. A radio lacking that capability would be less expensive and could allow resources to be channeled elsewhere. However, it would also be incapable of communicating with all levels of command or with the forces of the other military services.
Figure 4.
Cost of the Capacity Utilized per Ops Net, by Command Level

(In fiscal year 2003 dollars/bit per second/net)
Source: Congressional Budget Office.

Would a reallocation of the Army's projected resources be a viable option for mitigating the mismatch between bandwidth supply and demand projected in the future Objective Force? The results presented in the figures above do not offer clear guidance. Consequently, such a reallocation does not appear to be an obvious alternative for addressing the Army's future bandwidth bottleneck.

Reduce Demand and Better Manage Persisting Mismatches

Because of the potential problems associated with buying more bandwidth to eliminate the Army's projected bottleneck, CBO considered options to reduce lower-priority bandwidth needs and to better manage the remaining demand. Three such alternatives are discussed below. The first two would end the transmission of information that might be of lesser priority yet would be expected to contribute significantly (either directly or indirectly) to the future supply/demand mismatch at the higher command levels. The third option focuses on adopting software tools that are now entering the commercial market and that could allow better management of the demand for bandwidth when it exceeds supply. In keeping with its mission of providing impartial analysis, CBO makes no recommendations about adopting one or more of these options. Rather, they are meant to illustrate some of the choices that the Army might consider as it attempts to match its expectations regarding digitization to the communications capability it is buying.

Eliminate Video Teleconferencing on the Battlefield

As discussed earlier, video teleconferencing is a bandwidth-intensive activity; the usual teleconferencing site requires about 1,000 Kbps of bandwidth. Video teleconferencing enables commanders at disparate locations to interact as if they were personally present at one location. Participants can hear vocal tones and observe facial expressions and other aspects of behavior that commanders argue are important elements in decisionmaking and planning. In addition, they can observe and comment collectively on complex visual objects--for example, maps depicting the disposition of enemy and friendly forces.

In cases in which video teleconferencing capability does not exist or is unavailable because of a lack of bandwidth, an alternative is voice conferencing, which has been and continues to be used routinely by commanders as a substitute. Although voice conferencing does not allow conferees to collectively view objects or drawings, that capability can be provided by adding a "smart" whiteboard visual display; the digital whiteboard allows conferees to view a rendition (in black and white, not color, which would require more bandwidth) of the object under discussion--typically a dynamically generated drawing, map, or operational plan. (The conclusion that such a capability could be useful was reached by the 5th Corps just prior to the recent Iraq war.)(5) Conferees can not only use the whiteboard as a blackboard but may also utilize electronic features such as the ability to scan documents and communally view files.

Overall, voice conferencing coupled with whiteboard capability generally requires less bandwidth (by about a factor of 10) than video teleconferencing requires. The Army's test community has assessed the relative value, for collaborative planning, of teleconferencing, a drawing or black-and-white projection capability (available with the whiteboard), and video. Its judgment: "The draw and voice features deliver the highest value for collaboration. The draw feature was particularly critical. Video is a good feature but has the least relative value as compared to draw and voice capabilities."(6)

A second military use for video teleconferencing is telemedicine, in which medical personnel consult with doctors at remote locations to bring to bear experience and expertise that may not be available on the battlefield. Telemedicine has been a feature of recent low-intensity operations (such as the U.S. military's peace enforcement mission in Macedonia) during which only limited medical support was available in the theater of operations. But telemedicine's value in other military operations has been questioned. In-theater medical capability is extensive in high-intensity battlefield situations such as Operation Iraqi Freedom. Above the medical aid stations at the battalion level of command, the medical evacuation system delivers patients to combat-support hospitals and to field and general hospitals. Those large facilities have scores of physicians who are trained in battlefield and specialty medicine and who may not have the time or the need to consult extensively with off-site doctors. Hospital facilities are reinforced by an Air Force-run medical evacuation system that can move patients quickly to major medical centers in Europe, Japan, or the United States. After considering all of those factors, the Army has chosen not to provide telemedicine capability at the battalion level of command.

Eliminating video teleconferencing would reduce the current demand for bandwidth by as much as 70 percent at the brigade command level and 15 percent at the corps level, CBO estimates. Less demand at the brigade level would be helpful in reducing the bottleneck there, but the 15 percent drop in demand at the corps level would mitigate little of the bandwidth supply/demand mismatch there, where it is projected to be more severe in 2010. Those same reductions would apply to the demand for bandwidth in 2010 under the scenario in which video data collected by UAVs were not shared over the network. If such data were shared, which is a more likely outcome, then total bandwidth demand would be greater, and the fractional savings associated with eliminating video teleconferencing would be reduced to about 20 percent at the brigade level and 5 percent at the corps level.

Eliminate the Requirement to Support Networked UAVs

As discussed in Chapter 2, the demand for Army bandwidth in 2010 will exceed the projected supply by as much as a factor of 30 at the division and corps levels of command and up to a factor of 10 at the brigade level. Those estimates apply, regardless of the information management scheme that the Army decides to use, if the video data collected by UAVs are shared among operations networks at the upper levels of command. Under this option, the Army would forgo sharing those data beyond the initial sensor-to-shooter downlink. Instead, UAV operators would summarize collected intelligence in a lower-bandwidth message, which could then be more easily transmitted. Commanders would thus receive the results of UAV missions but would not have direct access, in real time, to the video images that the vehicles collected. Eliminating the requirement to share video data would reduce the future demand for bandwidth at the brigade and higher levels of command by a factor of three.

Provide Bandwidth Management Tools at the Applications Layer

If the Army eliminated both video teleconferencing and the requirement to share UAV data, CBO estimates that the service could reduce total bandwidth demand by roughly a factor of three. Nevertheless, such a reduction would be insufficient to bring demand into line with supply. Currently, when the supply of bandwidth is inadequate, operations officers attempt to manage demand manually as they transmit messages. They prioritize transmissions, on the basis of their experience and professional judgment, assigning a sequence to the times at which messages in a queue will be sent. Sometimes, to decrease delays and increase message completion rates, communications personnel in a tactical operations center literally pull the plug temporarily on some equipment in order to have enough bandwidth for the highest-priority messages. As discussed in Chapter 1, however, those efforts to reduce the demand for bandwidth have frequently failed to prevent unacceptable degradation in the performance of the Army's battlefield communications networks during experiments. During the recent war in Iraq as well, constraints on bandwidth forced soldiers to confront their lack of techniques for managing bandwidth.(7)

Communications officers have automated network management tools available to help them monitor a variety of measures of bandwidth demand and thus aid their prioritizing of message traffic. Some of those tools can automatically increase or decrease the bandwidth allocated to given types of messages or other data transmitted over the Army's battlefield network. But there is little or no feedback in such systems; consequently, the software applications that generate the messages and data transmitted over the network cannot automatically detect that the available bandwidth has been increased or decreased. As a result, message queues tend to lengthen once a bandwidth allocation has been reduced because the rate at which the messages are generated does not change. And once messages become old and exceed thresholds for latency (delay), they are eliminated from the network without being transmitted. In more extreme cases, the rates at which information flows between hardware components (which are usually set, by the flip of a physical switch, when the hardware is linked to the network) are exceeded, and messages are truncated or dropped. Neither condition is satisfactory: lost messages may be important, and overflowing queues, a portion of which the network continually tries to retransmit, strain the network's responsiveness and reliability.

This option would expand the capabilities of today's network management tools so that message-producing software applications could sense changes in the available bandwidth and automatically increase or decrease the rate at which messages were generated. That approach is beginning to be used by private-sector firms that develop software for use on the Internet.(8) The cost of developing and testing the software required to implement this kind of an option for the Army is currently unknown, and CBO did not attempt to estimate it.

1.  CBO analysts questioned program managers about the adequacy of funding levels as one way to gauge the severity of the risk that products scheduled for fielding by 2010 might not actually be available. Managers reported full funding, which signifies the Army's commitment throughout the Department of Defense's Future Years Defense Program to provide funds that the service and the Defense Department agree will meet the programs' formal requirements. For products scheduled for fielding after 2010 (for example, JTRS Cluster 2 cellular phones), program managers identified unfunded requirements that if fully funded might accelerate fielding dates. However, bringing those capabilities to the field faster would not mitigate the shortfalls that CBO's analysis revealed.
2.  See P. Gupta and P.R. Kumar, "The Capacity of Wireless Networks," IEEE Trans Information Theory, vol. 46, no. 2 (March 2000), pp. 388-404.
3.  The development and adoption of new methods--including so-called dynamic protocols--for transmitting information among the nodes in a communications network might reduce the inefficiency penalty that accrues as the number of nodes increases. However, the development and adoption of such protocols for use in military communications networks are unlikely to occur prior to 2010.
4.  The Army's estimates in 2002 of the total cost for the WIN-T program ranged from $4 billion to $9 billion (in 2002 dollars), with $2 billion in expenditures scheduled between 2004 and 2009. More recent (2003) estimates by contractors put total program costs at about $5 billion and $6.6 billion, respectively. The figures that follow use $5 billion. (Unless otherwise indicated, all costs are expressed as fiscal year 2003 dollars.)
5.  After several years of experience with video teleconferencing (VTC), the 5th Corps's commander, Lt. Gen. William Wallace, suggested in 2002 that VTC "was wasteful," and the corps went to war in Iraq with voice conferencing and a rudimentary whiteboard capability. A lack of training and established procedures prevented those tools from being judged fully successful, but the bandwidth advantage they offered (their bandwidth demand is much less than that of video teleconferencing) continues to spur additional experimentation. See Maj. M. Shaaber, Capt. S. Hedberg, and Troy Wesson, V Corps: C4ISR Integration AAR (after-action report) (May 2003), pp. 37-38.
6.  Lee Offen and Mary E. Stafford, Assessment Report for the Division XXI Advanced Warfighting Experiment (DAWE) (United States Army Operational Test and Evaluation Command, January 22, 1998), p. ES-11.
7.  Shaaber, Hedberg, and Wesson noted in V Corps: C4ISR Integration AAR (p. 5): "The ability to manage bandwidth usage dynamically at the discretion of the commander would [be helpful]."
8.  The Defense Information System Agency cites the reduction of bandwidth demand as one of the goals of its Defense Information Infrastructure Common Operating Environment (DII COE). The DII COE is a set of standards pertinent to the Defense Department's hardware, software, and networking capabilities. A number of commercial companies are beginning to respond with more-capable software tools that control computer file server software on the basis of the available network bandwidth.

Previous Page Table of Contents Next Page