Towards a DoD Spectrum Roadmap

In response to a letter from  the Senate Armed Services Committee to Secretary of Defense Lloyd Austin on June 2022, has DoD issued a “Request for Information to support the development of a Next-Generation Electromagnetic Spectrum (EMS) Strategic Roadmap.” This is my response.


At a high level, the RFI is grounded in the Department of Defense’s desire to achieve “freedom of action in spectrum at the time and place of its choosing.” The RFI declares that this goal decomposes into five interdependent sub-goals:

  1. Developing superior spectrum capabilities;
  2. Evolving to an agile, and fully integrated, spectrum infrastructure;
  3. Pursue total force readiness in the spectrum;
  4. Secure enduring partnerships for spectrum advantage; and
  5. Establish effective spectrum governance.

In particular, the RFI is interested in “revolutionary leap-ahead spectrum-based technologies and capabilities designed to share spectrum” as a means to the stated ends. Unfortunately, the RFI fails to cleanly separate operational ends and means in the domestic theater from those in other parts of the world. But much of value remains despite this miscue.

The Question of Context

While DoD may well be able to influence law and regulation in the US and, to a lesser extent, among NATO partners, it is unlikely to achieve substantial influence among the international standards bodies such as IETF, 3GPP, IEEE 802, Cable Labs, et al.

Furthermore, prospects for US influence over the technical infrastructure in peer and near-peer adversaries are limited. China in particular is capable of creating its own standards and manufacturing semiconductor chips and software to operationalize them. Adversaries are unlikely accept weak security even when it is produced by international standards organizations.

Hence, operating through in adversary states will require substantial technical innovation that takes installed communication infrastructure as it is. Guiding US technology along a path that foreswears interference or relocation of incumbent users (such as DoD) will not bring DoD closer to its ultimate goals.

Hence, the framing of the RFI interferes with DoD’s ability to gather the kind of information it needs to plot a roadmap for spectrum use in the ten to twenty year planning horizon upon which the Department operates.

While we will do the best we can to answer the RFI’s questions as written, we suggest that a further RFI is needed, one that focuses exclusively on the hostile theater and “hostile actor in a friendly theater” scenarios.

Three Elements of the Future of Warfare

For purposes of this inquiry, we assume that the future of warfare will consist of evolutions of existing and emerging elements as well as “black swans” that are currently unknown.

  1. Advanced Platform Development

On the existing technology front, the B-21 Raider is a good starting point that illustrates the value of modular design and advanced approaches to technology platforms:

The B-21 incorporates an open architecture, cloud computing, modular hardware, and agile software. These concepts are not everyday ideas, but suffice it to say they’re basic to the smartphones most of us carry. The key feature is the ability to upgrade the radio by unplugging the old one and plugging in a new one.[1]

These design concepts emerged in the civilian sector and have now appeared in military systems. Given the ten- to twenty-year DoD planning horizon, it’s vital for DoD to have the ability to adapt current systems to future needs and to do so quickly.

  1. Advanced Networks

LEO constellations are showing their value for warfare in Ukraine, where a purely civilian system, Starlink by SpaceX, connects both civilians and warfighters to the Internet and to each other. SpaceX’s erratic management also illustrates the pitfalls inherent in relying on a single-source commercial system for military needs.[2]

Anticipating these problems and opportunities, ARPA has launched the Blackjack program to explore LEO constellations better suited for warfare.[3] One significant output of Blackjack is the Arrow constellation created by the Airbus US Space and Defense company:

  • IODA, the In-Orbit Demonstration Service provided by Airbus with the European Space Agency, facilitates in-orbit validation of new satellite concepts and technology demonstration systems to prove and derisk your new mission in LEO.
  • In the Blackjack programme, Airbus will provide an architecture demonstration intended to show the military utility of global low-Earth orbit constellations and mesh networks of lower size, weight and cost. DARPA will use the ARROW satellite buses and pair them with sensors and payloads.[4]

It’s likely that LEO constellations will play an increasingly important role in future conflicts, but such networks have vulnerabilities. Hence, GEO satellites and ground-based wireless systems following future 3GPP standards and yet-to-be-created military approaches will still be necessary. 3GPP networks will become even more important than they are today.

  1. Artificial Intelligence and Machine Learning

The use of artificial intelligence (AI) and machine learning (ML) to create weapons with self-navigating and situational awareness capacity is beginning to emerge as both a near- and far-future option. One example of current AI analysis is the 2021 National Security Commission on Artificial Intelligence report. The report notes the defensive and offensive sides of AI.[5]

The use of AI in drones and other Internet of Things (IoT) devices is particularly important, as it has the potential to reduce the need for communicating navigation data and other information between drones and control centers on the ground and in the air.

While some elements of future warfare will place greater stress on networks, some elements, such as AI, have the potential to reduce network stress. AI/ML also plays a role in the management and upkeep of networks.[6]


In the following, I quote the RFI’s specific questions in italic and respond in roman (non-italic.)

A. What current or future technologies could facilitate harmonious spectrum sharing between unlike systems, such as next generation wireless and government radars?

 Technologies for effective spectrum sharing generally fall in the space of CDMA, SDMA, beamforming, subcarriers, MIMO, and traffic sensing. Further advances on this front will come from additional ways of manipulating waveforms and coding systems until each transmitter/receiver pair can create pathway of narrow beams carrying information that can only be seen, let alone decoded, by its intended recipient.

But the problem of spectrum sharing is more institutional than technical. Commercial networks struggle to share spectrum with government systems such as radar whose lineage precedes modern digital wireless networks by many decades. In many instances, the most efficient path to productive coexistence begins with upgrading the government system.

In the case of military radar, a radical redesign is well overdue. Military radar signals can and should masquerade as element of common commercial systems by using the same information-carrying waveforms that are ubiquitous in 3GPP, LEO, GEO, and Wi-Fi systems.

The use of such waveforms enables radar to function as a two-way communication network conveying, for example, Friend or Foe identification, frequency hopping patterns, and situational data.

In many theaters of operation, radar imagery can be extracted from actual 3GPP networks by integrating the receive side of the radar system into 3GPP base stations. Such implementations have been demonstrated by commercial vendors. By comparison with 3GPP networks, radar is at best a very rudimentary system occupying a large spectrum footprint to convey a relatively small amount of information.

A stark picture of radar inefficiency has been widely discussed in public media since November 2022. Commercial airliners, private planes, and helicopters utilize radar (or “radio”) altimeters that require hundreds of megahertz of spectrum to communicate one number, the altitude of the aviation vehicle.[7] The reliance of commercial aviation on obsolete and poorly designed altimeters prevents cellular networks from using spectrum rights they lawfully won at auction.

But the story is worse than this description conveys: testing by NTIA’s Institute of Telecommunication Sciences (ITS) at the Table Mountain quiet zone in Colorado indicates that the restrictions on radio signal strength and direction demanded by the aviation industry cannot solve any of aviation’s problems. As ITS describes their measurement results, 5G signals do not encroach on the frequency range assigned to aviation:

Across the frequency range of 4200–4400 MHz, we never saw any perceptible power from 5G emissions in the 100 ms time domain scans at each frequency step. We could visually see an impact of as little as 1/10 of a decibel above the analyzer’s noise. If such an emission had been present, we would have seen the very distinct TDD ON/OFF cycling every 5 ms. In effect, we would have seen “ripples” every 5 ms which would have been the “tops” of the TDD cycles shown in Figure 15.[8]

 In fact, harmful interference between well-behaved modern digital radio systems and legacy systems designed for operation in a vacuum can rarely be eliminated by the new entrant acting alone. The only effective solution is the wholesale replacement of the primitive system with a more capable upgrade.

The expression “harmonious spectrum sharing” brings to mind the recent experiment with CBRS. CBRS is an analog solution to a digital problem.[9] Rather than relying on Rube Goldberg systems for “sharing” spectrum hour-by-hour and day-by-day, the DoD should invest in the technologies listed above and in the most effective way to share spectrum that currently exists: becoming a customer of a commercial network that earns its daily bread by mediating the effective and efficient sharing of spectrum among its many users. The suggested solutions share spectrum millisecond-by-millisecond.

It is no longer necessary for every entity with a need for wireless communication to operate its own network.

B. What current and future technologies enable dynamic spectrum maneuverability and improve reliability in the use of the EMS in a congested and contested (i.e. adversary attempts to deny, disrupt, or disable spectrum-based capabilities) spectrum environments?

This question is quite similar to the previous one insofar as legacy systems often exhibit hostile characteristics to dissimilar systems. In fact, the common mode of disrupting wireless communication is jamming, simple transmission of an arbitrary, high power signal intended to saturate the adversary’s receiver, thereby rendering it deaf.

Similarly, the effective defense is rendering one’s own signal transparent to the adversary such that it doesn’t know what frequency to jam at what inclination. The best known past and present system is obviously frequency hopping spread spectrum, but it’s no longer interesting as originally formulated. The notion of agility is relevant, however.

In addition to the mechanisms already cited, present day smartphones provide a reasonable model for frequency agility, as they’re capable of using multiple licensed frequency bands, multiple unlicensed bands, and, recently, LEO satellite bands.[10] A DoD attempt at creating a similar capability, the Joint Tactical Radio System, wasted $6B without creating a usable system because its goals were too ambitious.[11]

In addition to a large inventory of terrestrial and satellite networks, ongoing work on waveform manipulation, modulation, impersonation, and piggybacking is worthy of examination. One idea that shows promise is the insertion of forged packets into commercial 3GPP network streams.

Using AI/ML, systems with information to transmit can predict idle periods in commercial networks that can be used to convey information to a knowing partner that the adversary can only detect with great difficulty. ARPA is already funding a project that goes along this line of inquiry.

C. What current or future technologies, tools, or platforms can improve DoD’s ability to handle large quantities of EMS environmental data to facilitate more dynamic forms of sharing including real-time and near-real-time spectrum sharing?

This is the AI/ML approach to opportunistic spectrum utilization. It depends on massive computation in a system at the network edge that can collect and sift packet data in order to predict time slots that are likely to be usable for packet insertion. At best, the predictions are speculative and the resource consumption is massive.

Such systems are only practical where the user has access to a compute facility with a steady supply of power. Hence, wartime use would be limited to domestic scenarios where carriers provide edge compute capabilities, allied nations with similar capabilities, and hostile theaters where the DoD has constructed such facilities on the ground or in the air.

Handling large volumes of data for the purpose of real-time prediction is energy and infrastructure intensive however it’s done. There is one approach that potentially solves the hostile theater scenario, however. Where LEO constellations exist, it’s possible to route data through the constellation from bird to bird, only dropping it to Earth outside the danger zone.

But this scenario raises an intriguing question: why would one choose to transfer information through a satellite constellation in support of a terrestrial network when the satellite network serves the purpose of massive data transfer all by itself?

D. What spectrum bands are candidates for sharing for commercial use to improve efficient use of electromagnetic spectrum?

We understand this question to imply “sharing by DoD with commercial networks”. Every band currently used by DoD and other government agencies is a candidate for sharing in one form or another. The most efficient sharing of spectrum suitable for commercial takes place when DoD becomes a customer of the commercial network (contingent, of course, on said network employing proper security.)

This understanding of the question puts the frequency bands already recognized by ITU and 3GPP at the head of the line: low band, mid-band, and mmWave band. The alternative sharing mode, CBRS, only allows for a periodic ping-pong pf control between DoD and the commercial operator. This mode of sharing is undesirable for reasons already noted.

An additional mode of “sharing”, loosely defined, follows the Eric Schmidt/PRC paradigm in which the government agency becomes the primary network user and operator, providing service to the civil sector on a commercial basis.[12] We have long doubted that this model is practical in the United States, and it’s certainly not at all practical in warfighting scenarios abroad.

E. What changes to legislative, regulatory, and policy frameworks could support increased non-federal spectrum access while simultaneously preserving DoD missions?

 This question searches for analog solutions to the problem of digital sharing. The offered venues are improper, and the question is poorly premised. Let’s return to the corresponding directive in the letter sent to DoD by the Senate Armed Services Committee before DoD paraphrased it, to wit:

(5) Potential gaps in processes and procedures within the Department to promote current and future EMS-based technologies for warfighter operations using federally allocated spectrum, while advancing national security and homeland defense missions;[13]

This is a more meaningful question as it gets to the institutional factors driving DoD to pursue a  quest for a spectrum sharing regime that does not exist today and probably will not exist in the future. We take “gaps in processes and procedures within the Department” as the operative phrase, and we don’t see any suggestion of changes to processes and procedure outside DoD.

One frequently mentioned gap is the federal contracting law that prevents outside parties from bargaining with DoD in “spectrum for networking” swaps. This is the scenario where a party, typically a network service provider, offers to upgrade a government system for no compensation except spectrum rights. The bargain allows the operator to upgrade a system, shrink its spectrum footprint, and then use the spectrum officially held by the government but no longer needed.

Such bargains would require changes in federal purchasing regulations. This is a path worth pursuing, but with considerable caution lest it create a patchwork of spectrum rights.

We’re more concerned about a certain mindset that seems to exist in the Pentagon. If DoD believes, as some critics charge, that only it can operate secure networks, we have a problem.[14] We also have a smaller, but more manageable, problem if institutional inertia and purchasing regulations conspire to make building proprietary DoD networks a more palatable choice than buying network services from competent providers, when such parties exist.

In the general course of things we would prefer for DoD to concentrate on its mission, leaving common products and services to contractors. DoD is a wonderful institution full of dedicated and talented staff, but the time has come when it no longer must be a network operator in addition to carrying out its many other responsibilities.

If it becomes apparent that DoD needs a LEO constellation as well as terrestrial networking, the path we would like to see the Department follow would entail joint specification with cooperating allies and operation by a stable and competent provider.  Each LEO constellation has global scope, so a partnership makes sense for its financial support as well.

F. How can DoD improve communications and long-term spectrum planning with the National Telecommunications and Information Administration on the military operational effect of electromagnetic spectrum policy decisions?

From the outside, it appears that DoD spectrum experts work well with ITS. It also appears that the present state of collaboration has more of a short-term focus than long-term one. Perhaps the creation of a formal long-term planning collaboration, along the lines of an office of strategic planning would be helpful. Locating DoD’s spectrum experts in the same general location as ITS (Boulder, Colorado) could well promote communication and collaboration as well.

G. Are there technologies provided by the telecommunication, information and communications technology industry, gaming/entertainment, or other areas outside traditional defense spectrum technologies that should be considered in a DoD EMS roadmap?

Gaming is a close predictor of the future of warfare, hence a collaboration could be beneficial. But certain culture barriers may be hard to overcome. Ties between DoD and academe are already strong but could be stronger. Certainly, ICT is core to the DoD mission.

H. How could cooperation and collaboration between DoD and external stakeholders be strengthened? How could cooperation and collaboration among federal agencies be strengthened?

DoD would be wise to cultivate peer relationships with contractors and other federal agencies. A number of programs exist to cultivate such relationships, of course. Much inter-agency collaboration depends on the sense of trust that is key to the Internet’s cooperative system. It is promoted by frequent face-to-face contact in informal settings. See NANOG for examples.

I. Are there any other matters that you would like to share with the Department of Defense to inform development of a threat-informed roadmap?

As General von Clausewitz may have said: “The enemy of a good plan is the dream of a perfect plan.” Strategic Roadmaps are often considered as little more than marketing documents in the ICT industry, so DoD would be wise not to fall in love with the output of this RFI.

A Silicon Valley aphorism asserts: “less happens in five years than expected, but much more happens in ten years.” The Next-Generation EMS Strategic Roadmap will be provisional, requiring revision at least midway through the next generation as the planning begins for the generation after that.

To the extent feasible, it’s wise to state goals and ambitions in such a way that they can be confirmed or disproved. Had this been done in the midst of the former “opportunistic spectrum sharing” bubble of the 2000s, much time and money could have been saved. Advanced spectrum sharing ultimately depends on the creation of tangible advanced spectrum sharing technologies that address wave form manipulation all the way into the quantum realm.

Amazing experiments have been done with quantum effects such as orbital angular momentum.[15] [16] Results are difficult to analyze, but this work should continue.


 As I wrote in the abstract of a paper for TPRC 41:

Spectrum allocation and management is an ongoing process that will benefit from guidance by a set of fundamental technical principles. Historically, spectrum allocation has been an ad hoc, piecemeal system driven by the logic of the moment: in most cases, a commercial enterprise or government agency with a need requested an allocation, and if the regulator agreed, it allocated the best available fit from the inventory. In other cases, spectrum assignment has been initiated by the regulator itself, either to good effect or otherwise.

The result of 80 years of ad hoc allocation is a system in which neighboring allocations sometimes pose tremendous burdens on each other, particularly in cases where high-power systems adjoin low-power ones. Such allocation errors give rise to intractable disputes over spectrum usage rights. Market dynamics are helpful, but not altogether sufficient to create a system of rational allocation as each player maximizes its own interests, which in the short term preserve inefficient allocations in the overall frequency map.

A more rational system of spectrum assignment would respect the principles that are evident in the operation of actual high-demand, high-performance, and high-efficiency wireless networks and in the trajectory of near-term spectrum research and development.[17]

 The quest for “principles evident in the operation of actual high-demand, high-performance, and high-efficiency wireless networks and in the trajectory of near-term spectrum research and development” continues. I welcome DoD to the search, even though you were in it before we were.


[1] Richard Bennett, “The End of ‘Airplane Mode,’” Washington Times, December 14, 2022,

[2] Matt Novak, “SpaceX Stops Ukraine’s Ability To Use Starlink Internet For Drones,” Forbes, accessed February 16, 2023,

[3] “Project Blackjack: DARPA’s LEO Satellites Take Off,” Airforce Technology, July 23, 2020,

[4] Airbus, “Arrow Brochure” (Airbus U.S. Space & Defense, Inc, 2020),

[5] “National Security Commission on Artificial Intelligence Final Report” (National Security Commission on Artificial Intelligence, October 5, 2021).

[6] “What Is Artificial Intelligence (AI) in Networking?,” Cisco Systems, Inc., accessed February 16, 2023,

[7] Richard Bennett, “Radio Amateur Hour,” High Tech Forum (blog), accessed February 16, 2023,

[8] Richard Bennett, “FAA Proposes Token 5G Fix,” High Tech Forum (blog), January 12, 2023,

[9] Richard Bennett, “Effective and Efficient Wireless Networks,” High Tech Forum (blog), September 20, 2022,

[10] Apple Support, “Use Emergency SOS via Satellite on Your IPhone 14,” corporate site, Apple Computer, 2023,

[11] Sean Gallagher, “How to Blow $6 Billion on a Tech Project,” Ars Technica (blog), June 18, 2012,

[12] Richard Bennett, “Eric Schmidt’s Spectrum Agenda,” High Tech Forum (blog), May 5, 2022,

[13] Committee on Armed Services United States Senate, “James M. Inhofe National Defense Authorization Act for Fiscal Year 2023, Report to Accompany S. 4543” (United States Senate, July 18, 2022).

[14] Sue Halpern, “The Terrifying Potential of 5G Technology,” The New Yorker, April 26, 2019,

[15] Wenchi Cheng et al., “Orbital Angular Momentum for Wireless Communications,” ArXiv:1804.07442v1, April 20, 2018.

[16] “Light Orbital Angular Momentum,” in Wikipedia, accessed June 29, 2012,

[17] Richard Bennett, “Technical Principles of Spectrum Allocation,” SSRN Scholarly Paper (Rochester, NY, 2013),