Handling the sensor deluge for spectrum superiority in CJADC2 operations

• Share on X
• Share on LinkedIn

Story

August 11, 2026

The U.S. Department of Defense (DoD) is undertaking one of the most ambitious modernization efforts in its history as it works to realize a unified Combined Joint All-Domain Command and Control (CJADC2) approach, under which sensors and systems are connected across the tactical, operational, and strategic levels to create clear situational awareness for the warfighter. Recent federal budget initiatives include more than $2 billion in the fiscal year 2027 request to transition joint command-and-control capabilities from fragmented legacy deployments into integrated operational programs, up from roughly $240 million the year before.

On paper, the shift toward artificial intelligence (AI)-enabled warfighting appears straightforward. The reality at the tactical edge is much different.

Today’s battlespace is generating more data than operators can realistically move, process, and act on in real time. Unmanned aerial systems (UASs), software-defined radios (SDRs), cellular arrays, and Internet of Military Things (IoMT) platforms continuously collect vast quantities of radio-frequency (RF) and signal data. The volume is only going to increase. Recent Gartner research¹ projects that by 2030, 90% of newly procured military hardware – including wearables, remote sensors, and weapons systems – will be integrated into an IoMT ecosystem. Much of that information must travel across communications links that were never designed to support this volume of traffic.

In contested and electronically degraded environments, operators often work with less than 2 Mb/sec of available bandwidth. When multigigabit spectral captures wait for satellite connectivity and centralized processing, decision cycles slow and electronic warfare (EW) targeting loops stall. The Common Opera­tional Picture becomes less accurate when speed matters most.

This is not simply a data-management challenge but is actually a spectrum-superiority challenge. The Gartner research noted that, “The overwhelming proliferation of data from a vast amount of battlefield devices, connected through the IoMT, will create a new warfighting challenge.”

If CJADC2 is going to deliver its intended advantage, military architectures must move beyond centralized processing models and place intelligence closer to the point of collection. The services are already moving in this direction. In June 2026, the Army placed a $350 million order to field a system that senses and reports a command post’s electromagnetic-spectrum signature in near-real-time, an explicit bet on pushing spectrum awareness to the tactical edge.

Empowering the edge instead of the enterprise cloud

For more than a decade, defense AI development has largely followed the commercial cloud-computing model, under the assumption that larger computing clusters can solve increasingly complex processing problems. That assumption does not hold up at the tactical edge.

Modern electromagnetic warfare is built around disruption. Adversaries will target networks, jam communications, and degrade connectivity whenever possible, making backhaul of information impossible. The situation on the ground in Ukraine has made this concrete. With tens of thousands of jammers now lining the front, both sides have turned to jam-resistant autonomy, including neural-network optical navigation and fiber-optic guided drones that bypass radio-frequency and satellite links entirely. Any system that depends on a constant connection to a CONUS-based [contiguous 48 U.S. states] cloud environment introduces operational risk.

Building a resilient force requires systems that can function under denied, degraded, intermittent, and limited (known as DDIL) conditions. Recent Pentagon AI initiatives have consistently emphasized moving data-centric operations closer to the mission, as people in the field are asking for this directly. At SOF Week 2026, special-operations leaders described wanting data center-class AI capability on disconnected front lines, naming decision support, intelligence analysis, and mission planning as priority applications. Many decisions need to happen where data is collected rather than after it reaches a distant operation or data center.

The defense community still underestimates the importance of this shift. Many organizations continue to assume that larger architectures automatically produce better outcomes. In bandwidth-constrained environments, efficiency matters more than scale and spectrum superiority depends on placing processing power where the data originates.

The side that can identify, classify, and act on signals first gains a measurable operational advantage. That advantage is difficult to achieve when every sensor depends on a distant cloud environment to make sense of what it sees.

Embedding AI directly into tactical systems

The most practical way to eliminate the backhaul bottleneck is to process data at the sensor level.

Instead of transmitting raw RF spectrum imagery and IQ [in-phase and quadrature] data to remote infrastructure, embedded systems can run optimized machine learning (ML) models locally. Those models can perform signal-signature recognition, pattern-of-life analysis, and threat identification in real time. The goal is to move only the required information rather than move more data across the network.

Achieving that outcome requires modular, multi-node sensor architectures capable of generating trusted alerts without saturating communications links. Modern edge AI makes this increasingly practical. Advances in quantization, pruning, and hardware-specific compilation enable highly specialized neural networks to run efficiently on low-power embedded processors.

These systems can analyze millions of spectral samples every second. They can identify anomalous transmissions, classify radar emissions, and detect cellular signatures directly on the device collecting the data.

This approach aligns closely with recent Department of the Air Force AI priorities, which focus on fielding capabilities that solve operational problems at mission speed. The department is funding the connective tissue to match. A June 2026 Air Force award to build the Advanced Battle Management System (ABMS) digital infrastructure, valued at about $192 million, is aimed explicitly at accelerating CJADC2 and pushing AI deployment to the front line. Local processing can remove more than 90% of irrelevant spectral activity before information is transmitted across the network.

Rather than flooding communications channels with raw IQ data, edge systems can distribute only the information that matters. Emitter coordinates, threat classifications, and priority alerts consume a fraction of the bandwidth while providing substantially greater operational value.

For operators tasked with maintaining awareness across increasingly congested portions of the electromagnetic spectrum, that reduction in data volume is more than an efficiency gain, but it is actually what makes spectrum superiority achievable under real-world operational conditions.

Achieving subsecond response times

Speed is the primary advantage of edge-native AI. In modern electromagnetic warfare, adversary radar and jamming systems often transmit for only fractions of a second to avoid detection and anti-radiation targeting. Traditional workflows struggle to keep pace. Data must be collected, buffered, transmitted over constrained communications links, processed remotely, and returned to operators. That process can take minutes, which is often too long.

By processing information locally, embedded AI systems can identify threats and initiate responses in less than a second. A detected waveform can immediately trigger electronic counter­measures, cue kinetic-strike assets, or distribute alerts across the force using compact data packets.

The result is a faster sensor-to-shooter cycle and a shorter path from detection to action. In conflict, that speed advantage can determine whether an opportunity is exploited or lost. Industry is already collapsing these timelines: Commercial counter-drone systems now advertise fusing detection and defeat to cut decision time from minutes to seconds, as documented very recently when an advanced defense-technology company selected an AI-powered command-and-control layer for its counter-UAS capability.

More importantly, edge-native AI can enable commanders to maintain awareness and decision advantage across the electromagnetic spectrum while adversaries are still processing what happened.

Building the next generation of joint systems

Achieving this vision will require close collaboration between software devel­opers, defense integrators, and em­bedded hardware manufacturers.

Future systems must support containerized applications, modular software workloads, and open standards such as the Sensor Open Systems Architecture, or SOSA, approach. These design choices make it possible to update algorithms in the field without replacing hardware.

Investment priorities should focus on software and hardware codesign. Success will come from extracting maximum performance from constrained edge devices while maintaining flexibility for future mission requirements. That investment is already mobilizing. One recent market study saw analysts projecting that EW spending will climb from $15.62 billion in 2026 to $23.85 billion by 2031, with spectrum-dominance and counter-drone programs cited as the primary growth drivers.

Notes: ¹ Gartner, Top Trends in Defense for 2025: Internet of Military Things, www.gartner.com/en/documents/6941366, Jay Phipps et. al, 10 September 2025

Dr. Michael P. Jenkins is the chief product and technology officer at Knowmadics, where he leads product strategy and technology development across the company’s sensing, situational-awareness, and AI-enabled software and analytics portfolio. A cognitive systems engineer with more than 20 years of experience, he specializes in human-machine teaming and the design of decision-support systems for national-security partners.

Knowmadics • https://knowmadics.com/