The Internet of Battle Things

Alexander Kott, U.S. Army Research Laboratory

Ananthram Swami, Fellow IEEE, U.S. Army Research Laboratory

Bruce J. West, Fellow AAAS and APS, U.S. Army Research Office

[This is a version of the article that appears in IEEE Computer as:

Kott, Alexander, Ananthram Swami, and Bruce J. West. "The Internet of Battle Things."
Computer 49.12 (2016): 70-75.]

The rapid emergence of Internet of Things is propelled by the logic of two irresistible
technological arguments: machine intelligence and networked communications. Things are more
useful and effective when they are smarter, and even more so when they can talk to each other.
Exactly the same logic applies to things that populate the world of military battles. They too can
serve the human warfighters better when they possess more intelligence and more ways to
coordinate their actions among themselves. We call this the Internet of Battle Things, IoBT. In
some ways, IoBT is already becoming a reality1, but 20-30 years from now it is likely to become
a dominant presence in warfare.

The battlefield of the future will be densely populated by a variety of entities (“things”) – some
intelligent and some only marginally so – performing a broad range of tasks: sensing,
communicating, acting, and collaborating with each other and human warfighters2. They will
include sensors, munitions, weapons, vehicles, robots, and human-wearable devices. Their
capabilities will include selectively collecting and processing information, acting as agents to
support sensemaking, undertaking coordinated defensive actions, and unleashing a variety of
effects on the adversary. They will do all this collaboratively, continually communicating,
coordinating, negotiating and jointly planning and executing their activities. In other words, they
will be the Internet of Battle Things.

1
Figure 1 Broad variety of systems and other "things" will communicate and collaborate on
the battlefield. (Source: Illustration by Evan Jensen, U.S. Army Research Laboratory)
To become a reality, however, this bold vision will have to overcome a number of major
challenges. As one example of such a challenge, the communications among things will have to
be flexible and adaptive to rapidly changing situations and military missions. This will involve
organizing and managing large number of dynamic assets (devices and channels) to achieve
changing objectives with multiple complex tradeoffs. Such adaptation, management and re-
organization of the networks must be accomplished almost entirely autonomously, in order to
avoid imposing additional burdens on the human warfighters, and without much reliance on
support and maintenance services. How can this be done?

Secondly, human warfighters, under extreme cognitive and physical stress, will be strongly
challenged by the massive complexity of the IoBT and of the information it will produce and

2
carry. IoBT will have to assist the humans in making useful sense of this massive, complex,
confusing, and potentially deceptive ocean of information, while taking into account the ever-
changing mission, as well as, the social, cognitive and physical needs of humans.

Finally, nobody can discount the most important feature of the battle – the enemy. Besides being
a lethal physical threat to the humans and IoBT, the enemy will be lurking in and around the
IoBT networks and its information. IoBT itself will be a battlefield between its owners and
defenders, and its uninvited part-owners – attackers. How will IoBT manage risk and uncertainty
in this highly adversarial, deceptive environment?

These are some of the questions that were discussed at the strategic planning meeting that was
organized by the US Army Research Laboratory (http://www.arl.army.mil) on 9-10 November
2015, and brought together a number of scientists from academia and industry, and military
experts. The suggestions and concerns that emerged at the meeting coalesced into a rich and
ambitious research agenda, summarized below.

Managing and Adapting the IoBT

In spite of voluminous, current and past research on related topics in network science and
engineering, merely by virtue of its exceptionally large scale IoBT will require new theoretical
results, models, concepts and technical approaches. Indeed, IoBT’s number of nodes for a future
Army brigade might be several orders of magnitude greater than anything that has been
considered in current practice. This is particularly true in the environments where such a brigade
will find it advantageous to make use of networked devices and channels that it does not own,
e.g., when making use of existing local civilian IoT (networking infrastructure and things) in
military operations in a megacity. In this case, the meeting’s participants suggested, IoBT scale
on the order of a million of things per square kilometer is not an unreasonable target for
exploration.

3
Figure 2 Combatants will perform cyber attacks partly through the civilian Internet of
Things to which they will be inevitably connected. (Source: Illustration by Evan Jensen,
U.S. Army Research Laboratory)
On the other hand, the massive scale of IoBT can be advantageous in practice and even for
theoretical purposes. For example, availability of very large and densely positioned number of
things, such as sensors can help eliminate currently common concerns about availability of any
of them at a given time. To this end, theoretical results are needed to understand the degree of
determinism resulting from very large ensemble of things and data.

Quite apart from its large scale, extreme heterogeneity of IoBT will call for new research and
approaches. Not only the local IoT will consist of a broad range of commercial things and
networks, but even the equipment that the warfighters will bring with them into the battle will
likely rely on commercial offerings. It is probable that future commercial IoT will continue to
exhibit a lack of standards, partly driven by desire of individual manufacturers to control its
market, and will be generally chaotic. The military will have to adapt rapidly – and to have
suitable technologies and techniques for such an adaptation – to use a broad variety of things,
protocols, and communication technologies from multiple manufacturers.

In such a heterogeneous, highly dynamic and largely unpredictable environment, new

approaches will be needed to facilitate discovery, characterization and tracking of relevant,
available and useful things, dynamically in time and space. In particular, a military force that

4
utilizes an existing IoT of a local society, e.g., a megacity, will not able to make reliable
assumptions about behaviors and performance characteristics of any parts of its IoBT; instead
such behaviors and characteristics will have to be learned and updated automatically and
dynamically during the operation. Speaking of complex and unpredictable behaviors, one must
not forget that humans – whether we call them “things” or not – are crucial and highly influential
elements of IoBT. Behaviors and intents of humans – friendly warfighters, adversaries, and
neutral civilians – will have to be dynamically detected, identified, characterized and projected in
order to operate the IoBT.

Communications between things will also be challenged by high complexity, the dynamics and
the scale of IoBT. Finding, sharing and managing communication channels, between large
numbers of competing, heterogeneous and often unpredictable things will require novel
approaches. Highly intelligent automation will be required to continually allocate and
reconfigure the resources of the communications network. Information-sharing strategies and
policies – who talks to whom, when, about what, and how long – will have to be automatically
designed and modified dynamically. Highly scalable architectures and protocols will be needed,
along with rigorous methods to determine and validate properties of protocols and architectures.
In extreme situations, when IoBT experiences catastrophic collapse or becomes largely
unavailable, or untrustworthy due to enemy actions, the autonomous management of IoBT will
need to provide at least a “get me home” capability, which will enable the continuation of
operations , even if at a limited level of functionality.

Additional complexity will arise from the wide range of timing constraints on communications.
Some communications can wait for hours, while other communications will pose real-time
requirements, for example for sensing and actuating. The channels will be constrained in highly
heterogeneous ways as well. It is expected that 30 years from now, consumers will use wireless
channels typically for only a few meters before the data enters fiber or other high-capacity
channels; at the same time the military will require at least a few kilometers of wireless channels
before encountering fiber.

To enable the dynamic management of IoBT, situational awareness of the IoBT as a whole will
be formulated and updated rapidly and automatically; therefore new approaches will be desired
directed towards the ability to measure relatively few variables of the complex system while
thereby obtaining or inferring sufficiently complete information about the system.

While managing the IoBT, its purposes and uses must be taken into account, and these will be
diverse. Some of its purposes will be relatively well understood, such as tactical military logistics

5
or distributed computing. Others will be novel and will emerge from the availability of IoBT
itself, such as perhaps use of IoBT for Position, Navigation and Timing (PNT) needs, and as a
supplement to, or replacement for, GPS.

Making IoBT Information Useful

As important as communications bandwidth is for effective operation of IoBT, it is the human
cognition bandwidth that will emerge as the most severe constraint3. Human warfighters do not
need and cannot process the enormously large flows of information produced and delivered by
IoBT. Instead, humans seek well-formed, reasonably-sized, essential information that is highly
relevant to their cognitive needs, such as effective indications and warnings that pertain to their
current situation and mission. Responding to each thing that demands the human’s attention, and
to each piece of data that seem vaguely interesting, is not a feasible option in the context of
IoBT. In fact, a key risk of IoBT is providing human warfighters (and intelligent things) with
inappropriate information that leads – or misleads – to an action with an outcome worse than
what would occur without that ‘information’.

To make its information useful, IoBT technologies will have to deal with a large volume and
complexity of information that are truly unprecedented in their extent4. Arguably, the quantity of
data within IoBT will far exceed any likely advances predicted by Moore’s Law and ever more
efficient use of bandwidth might offer in the future. Besides the sheer volume, the complexity of
the information will be formidable. For example, levels of abstraction, trustworthiness and value
of information (produced or consumed) will vary drastically between different things.

The very foundations of information theory will need to be reconsidered; for example, ensemble
probability densities are foundational for information theory, and require the underlying process
to be ergodic. However, the IoBT is expected to have nonlinear dynamic processes that are
sufficiently complex to generate events with non-ergodic statistics. The information entailed by
the occurrence of such events must be based on single time series and not on an ensemble of time
series5. Furthermore, non-intuitive, novel phenomena may emerge in the transfer of information
between dissimilar large networks. An example would be in how situational awareness is
modified by the information exchanged back and forth between IoBT and the social network of
human warfighters, see e.g., 6 and references therein. Such unexpected phenomena may also
influence – in yet-unknown ways – the ability of warfighters to control, inform and be informed
by IoBT.

6
Still, at the very least, the IoBT’s colossal volume of information must be reduced to a
manageable level, and to a reasonably meaningful content, before it is delivered to humans and
intelligent things. A likely target for compression and fusion of data into information, the
meeting’s participants conjectured, would be by a factor of 10E15. One approach to such a
challenging fusion task is to populate IoBT, with a layered hierarchy of information brokers7, or
“concierges”, which would aggregate, fuse, interpret and deliver appropriate information. The
fusion process should begin at the lowest possible level; for example, whenever possible, all
information-producing things should be equipped with the means to perform locally a degree of
filtering, interpretation and fusion, before sending data to the network. Although such layers of
intermediaries do complicate or restrict the discovery of underlying data, it may be a necessary
price to pay for arriving at useful, manageable and meaningful information.

However, for information brokers to do their job, they need to know what constitutes useful
information. Where would such knowledge come from? One source could be mission planning
and rehearsal that could help determine what information is required by the mission-performing
agents (human and artificial), and what is the likely available information. To capture the
resulting knowledge, a machine-interpretable, formal, broadly applicable and military-relevant
language will be needed for expressing information needs in highly heterogeneous IoBT8.
Moving beyond the inevitable limitations of mission planning and rehearsal, IoBT will need
approaches to self-learning of what information is needed for particular warfighter(s) and
particular mission. Such approaches will likely require a form of integration of machine learning
and semantic knowledge-based techniques.

More generally, executable models of the IoBT and its surrounding world are needed to enable
validation, interpretation, fusion, and assessing trustworthiness of the information (e.g., 9). Large
scale simulation may help large scale sensing and interpretation of information in a targeted,
purposeful manner. The research on formulating and automatically creating (and dynamically
maintaining) such models is in its infancy. Effective solutions to this challenge will likely
involve distributed self-modeling, self-calibration, and self-validation of IoBT.

Dealing with Deception and Adversarial Nature of IoBT

Nothing differentiates IoBT from IoT more than the battle – the B in IoBT – against a
determined and lethal enemy. The adversarial nature of the environment is the primary concern
in the life of IoBT. The enemy threatens physical survival and functioning of IoBT by kinetic,
directed-energy and electronic attacks against its things, by jamming the RF channels, by
destroying fiber channels and by depriving IoBT of its power sources. The enemy also threatens

7
the confidentiality, integrity, availability of the information within IoBT, by electronic
eavesdropping, and by deploying malware into IoBT10. Finally, and perhaps most importantly,
the enemy attacks the cognition of human warfighters. Humans will be elements within IoBT
that are most susceptible to deceptions, particularly to those based on cognitive and cultural
biases11. Humans’ use of IoBT will be handicapped when they are concerned (even if
incorrectly) that the information is untrustworthy12 or that some elements of IoBT are controlled
by the enemy. Similar susceptibilities, in part, apply to artificial intelligent entities.

Among the top priorities will be to minimize the enemy’s opportunities to acquire information
about IoBT and the warfighters it serves. While many of the applicable measures are the same as
those for conventional battlefield networks, the exceptional scale, heterogeneity and density of
IoBT offer additional opportunities for friendly information protection13. The sheer quantity of
things (especially in those cases when friendly forces leverage the local IoT) permits the use of
“disposable” security: devices that are believed to be potentially compromised by the enemy are
simply discarded or disconnected from the IoBT. To defeat the enemy’s eavesdropping, the
defenders may want to take advantage of plentiful availability of things and inject misleading
information into a fraction of them14. The density, complexity and diversity of message traffic
within the IoBT will make it more difficult for the enemy to perform the traditional traffic
analysis that could reveals details of the friendly command and control structure. Similarly, with
a large number and density of things, it may be less expensive and more efficient to stymie the
enemy’s cyber intrusions by creating large, believable honeypots and honeynets, which are
currently expensive to produce and to maintain dynamically. Although in the long run a
honeynet may be less expensive than the devastation wrought by an adversary’s cyber intrusion.

Besides acquiring friendly information, i.e., violating its confidentially, the enemy will attempt
to violate the information’s integrity, by modifying it with cyber malware, inserting rogue things
into IoBT, intercepting and corrupting it while in motion, between the things, and presenting
wrong information to the information-acquiring things, e.g., sensors. IoBT will likely fight back
by anomaly detection that can highlight unexpected data patterns, unexplained dynamic changes,
or lack of expected events (the dog that does not bark)15. To enable the anomaly detection,
machine learning approaches will be developed to deal with the data as big and as dynamic as
IoBT will possess. Such a continuous learning process will be computationally and bandwidth-
wise expensive. It will be further challenged by the possibility that the enemy will adapt and
evolve faster than the learning process can. In order to prevent the enemy from acquiring
physical or software modifications of friendly things, approaches will be needed to achieve
large-scale physical fingerprinting (e.g., collection of power consumption patterns) of things and
continuous IoBT-wide monitoring of such patterns16. More generally, there will be means for

8
active “stimulative intelligence” – ongoing physical and informational probing of IoBT that
could help reveal the structure and behavior, including anomalous and suspicious ones, of the
IoBT17.

Learning normal patterns and detecting anomalous deviations, however, does not work well
against a well-designed deception18, 19. In fact, learning can be a very dangerous double-edged
sword with respect to deception. A common approach to deception is for the enemy to cause the
friendly forces to learn a certain normal pattern, and then perform actions that blend into that
pattern, but result in an unanticipated outcome. Any measure of normalcy can be defeated by
effective deception. Still, very large scale and heterogeneity of IoBT may help defeat deception
because “lying consistently is difficult;” it may be particularly difficult when the available
sources of information are so numerous and are as heterogeneous as in IoBT. In general, much
research is needed on approaches to counter-deception, discovery or rejection of deception for
uniquely complex environment of IoBT25. And, considering that friendly IoBT will be
necessarily connected with the local civilian IoT and thereby to the enemy’s IoBT, approaches
are needed to performing offensive operations executed within the intertwined space of friendly
and enemy networks.

Such advanced capabilities will not be possible without new theoretical explorations. Fighting
the battle of IoBT may require major new results in game theory, particularly focused on
problems with very large number and very diverse game moves; near-infinite opportunities for
probing; high complexity of utility functions, and partial observability of the game board limited
to a very small fraction of the overall space. New theory is needed to formalize and normalize
diverse definitions and conceptualizations of risks20 and uncertainty. Deception should be
integral to this theoretical analysis. For example, theoretical results should help predict the
appropriate (or counterproductive) degree of complexity for a successful deception.

Disclaimer
This article does not reflect the positions or views of the authors’ employers.

References

9
1. George I. Seffers, “Defense Department Awakens to Internet of Things,” Signal
Magazine, January 1, 2015

2. Kott, Alexander, David S. Alberts, and Cliff Wang. "Will Cybersecurity Dictate the
Outcome of Future Wars?." Computer 48.12 (2015): 98-101.

3. Kranz, Matthias, Paul Holleis, and Albrecht Schmidt. "Embedded interaction: Interacting
with the internet of things." Internet Computing, IEEE 14.2 (2010): 46-53.

4. Miorandi, Daniele, et al. "Internet of things: Vision, applications and research

challenges." Ad Hoc Networks 10.7 (2012): 1497-1516.

5. B.J. West and P. Grigolini, Complex Webs; Anticipating the Improbable, Cambridge
University Press, Cambridge, UK (2011).

6. B.J. West, M. Turalska and P. Grigolini, Network of Echoes; Imitation, Innovation and
Invisible Leaders, Springer, NY (2014).

7. Korzun, Dmitry G., Sergey I. Balandin, and Andrei V. Gurtov. "Deployment of Smart
Spaces in Internet of Things: Overview of the design challenges." Internet of Things,
Smart Spaces, and Next Generation Networking. Springer Berlin Heidelberg, 2013. 48-
59.

8. Barnaghi, Payam, et al. "Semantics for the Internet of Things: early progress and back to
the future." International Journal on Semantic Web and Information Systems (IJSWIS)
8.1 (2012): 1-21.

9. Cho, Jin-Hee, Ananthram Swami, and Ing-Ray Chen. "A survey on trust management for
mobile ad hoc networks." Communications Surveys & Tutorials, IEEE 13.4 (2011): 562-
583.

10. Babar, Sachin, et al. "Proposed security model and threat taxonomy for the internet of
things (IoT)." Recent Trends in Network Security and Applications. Springer Berlin
Heidelberg, 2010. 420-429.

11. Kristin E. Heckman and Frank J. Stech, Cyber Denial, Deception and Counter Deception:
A Framework for Supporting Active Cyber Defense. Springer 2015

10
12. Yan, Zheng, Peng Zhang, and Athanasios V. Vasilakos. "A survey on trust management
for Internet of Things." Journal of network and computer applications 42 (2014): 120-
134.

13. Satyajayant Misra, Reza Tourani and Nahid Ebrahimi Majd, "Secure content delivery in
information-centric networks: design, implementation, and analyses," ACM SIGCOMM
Information-Centric Networking Workshop, pp. 73–78, 2012.

14. Bisdikian, Chatschik, et al. "Trust and obfuscation principles for quality of information in
emerging pervasive environments." Pervasive Computing and Communications
Workshops (PERCOM Workshops), 2012 IEEE International Conference on. IEEE,
2012.

15. Raza, Shahid, Linus Wallgren, and Thiemo Voigt. "SVELTE: Real-time intrusion
detection in the Internet of Things." Ad hoc networks 11.8 (2013): 2661-2674.

16. Roman, Rodrigo, Pablo Najera, and Javier Lopez. "Securing the internet of things."
Computer 44.9 (2011): 51-58.

17. Ashraf, Qazi Mamoon, and Mohamed Hadi Habaebi. "Autonomic schemes for threat
mitigation in Internet of Things." Journal of Network and Computer Applications 49
(2015): 112-127.

18. Vidalis, Stilianos, and Olga Angelopoulou. "Assessing identity theft in the Internet of
Things." IT CoNVergence PRActice (INPRA), 2 (1) (2014).

19. Teixeira, André, et al. "A cyber security study of a SCADA energy management system:
Stealthy deception attacks on the state estimator." arXiv preprint arXiv:1011.1828
(2010).

20. Kott, Alexander, and Curtis Arnold. "The promises and challenges of continuous
monitoring and risk scoring." Security & Privacy, IEEE 11.1 (2013): 90-93.

11