Sponsored by

Most 6G related articles treat Integrated Sensing and Communications as a future standard item (2030+): add radar-like functions to the same air interface that already serves phones. That is a reasonable commercial path. But it will not cut it for imminent mission-critical needs, such as defense for drone swarms. See example: https://youtu.be/rAXIsIwrtKI

OFDM was the waveform selected for 4G LTE more than 15 years ago. It offered superior performance and economics, while also providing the granularity to allow IP traffic to flow smoothly over cellular networks. This was demonstrated by Flarion Technologies as early as 2001 (acquired by QUALCOMM in 2006). Up until this point, carriers deploying 3G had to make do with the same “all you can eat” voice revenues despite spending billions for new 3G spectrum and billions more on deploying new 3G networks. This is why 3G had the shortest lifespan of any cellular generation, it fell short generating new revenue streams. Almost overnight, 4G ushered in the era of smartphones as we know today. This delivered carriers the incremental revenues needed to justify continued network investment and contributed to a decade of profitable growth.

Given the success of 4G, 3GPP architects selected OFDM again for 5G, redesigning it with wider channels to support network slicing, ultra-low latency, ultra-high reliability, and other features. But by simply using the same waveform with wider channels, the industry has again found itself in exactly the situation it faced with 3G.

Only OTFS can provide reliable, affordable wireless communications while enabling ISAC. It was “born” in the radar domain. 

• OTFS doesn’t require a cyclic prefix that causes OFDM performance to break down at high levels of mobility, especially at higher frequencies.
• OTFS has the capacity to exploit full-time-frequency (TF) diversity as well as the ability to work well in high Doppler fast-fading wireless channels. 

Cohere Technologies and partners are building a multi-waveform base station to pave the path for the U.S. government. In July 2026 the FutureG Office in the U.S. Department of War funded a $28 million prototype known formally as “Procurement of Multi-Waveform Radio Access Network (RAN) for Integrated Sensing and Communications (ISAC)” so that one open RAN stack can run ordinary 4G/5G OFDM and Pulsone Technology (Zak-OTFS) side by side. The sensing mission comes first. The existing cellular plant is the carrier, not the constraint.

Mission Critical

Tom Rondeau, Principal Director for FutureG, OUSW(R&E), did not describe a research curiosity. He described an urgent schedule. “ISAC is a mission-first priority for the U.S. Department of War to defend against drone swarms. Due to guidance from leadership to execute rapidly, we required a partner with the right technology ready today.” Cohere, he said, “offered a clear path that we could move on immediately,” because OTFS “carries information directly in the sensing domain” and because the work can land on the Linux Foundation’s OCUDU platform (a FutureG sponsored initiative).

That pairing of waveform plus open stack is the article’s real subject, a software-defined physical layer that can keep talking to today’s devices while a second waveform does the radar-like work OFDM does only with extra pilots, split resources, and a tax on capacity.

Dr. Anton Monk, Cohere’s SVP of Strategy, stated the operating rule: operators and national defense forces should be free “to choose the best waveform for the application, performance, and mission.” Mission-first ISAC is that rule applied to counter-UAS, border security, battlefield awareness, and critical infrastructure, not a demand that every consumer handset change modulation next year.

Why OFDM stays in the picture

Cohere does not argue that communication-based sensing is impossible on OFDM. It argues that “workable” is the wrong bar when a missed track is a missed threat. Time-frequency waveforms were optimized for bits. Sensing on them usually competes with those bits. Zak-OTFS was defined in Delay-Doppler space: delay is range and multipath; Doppler is velocity. The communication signal is the illuminator. There is no separate sensing allocation and no designed-in fight between the two functions.

That is why the prototype is multi-waveform rather than OTFS-only. OFDM remains the lingua franca of the installed base. Pulsone is the sensing-native option on the same radios, the same towers, and the same commercial bands—so the activity is hard to tell from ordinary cellular traffic. Ray Dolan, Cohere’s chairman and CEO, called the combination “affordable, scalable, and operationally invisible.”

Invisibility here is an operational requirement, not marketing. Concentrated radars are the first things taken down in a fight. A sensing function spread across a cellular grid can lose sites and still see. The same logic is why the program lists providing the targeting/handoff data needed to defeat drone threats as an output, not only pretty range-Doppler plots.

What has to be built

The physical layer must host OFDM and Zak-OTFS on an open, extensible stack. A mobile testbed must support bi-static and multi-static geometries, not only a co-located transmitter and receiver. Inference software must turn Delay-Doppler returns into 3D tracks with class labels and confidence. Outdoor ranges must exercise those modes. The whole thing must meet FutureG OCUDU and Zero Trust rules.

OCUDU is doing real architectural work. Cohere describes the award as integrating Zak-OTFS, multi-RAT spectrum sharing, and a neural receiver into the Linux Foundation OCUDU RAN stack. That is how a government prototype becomes something an operator can actually load, rather than a one-off radio stack in a lab.

Behind that stack sit products already aimed at commercial 4G and 5G. Universal Spectrum Multiplier software improves spatial multiplexing on networks that exist today. ECHO, layered on USM, maps the geometry of the radio channel and feeds bi-static and multi-static ISAC on current base stations. The government program does not invent a greenfield network. It puts a mission payload on a Delay-Doppler view of towers the country already paid for.

The receiver is part of the mission

A second waveform is useless if the receiver cannot demodulate it in real time next to OFDM. Cohere, Virginia Tech, and Duke showed a neural receiver for Zak-OTFS ISAC on NVIDIA’s Jetson AGX Orin at GTC in Washington. Dr. Lingjia Liu of Virginia Tech described is as extremely low complexity with no need for offline training. Dr. Robert Calderbank of Duke has pointed to Delay-Doppler as the reason the waveform is stable, predictable, and able to sense, the same properties that matter for high Doppler on the ground and for non-terrestrial links.

Performance figures tied to Zak-OTFS sensing—about 4× finer resolution, 4× more simultaneous targets, objects on the order of one-quarter the size OFDM-based ISAC would catch—are the reason Cohere treats waveform choice as a mission variable. Downstream AI costs are similar either way. What changes is what the front end ever sees.

Dual use without pretending the buyer is a phone company

The award still names commercial cousins: Advanced Air Mobility, traffic management, and public safety. Dolan called the contract validation of “sovereign, future-proof wireless infrastructure that serves both national security and commercial markets.” That is the dual-use sentence. It should not be read as the defense requirement being a 5G feature flag.

FutureG asked for a partner that could move now, on open infrastructure, with a waveform that already lives in the sensing domain. Cohere is delivering a multi-waveform RAN so the United States can keep the network it has and still field ISAC that is specified as mission-first: detect, classify, track, cue, and do it in bands that look like ordinary cellular while always watching for something in the air that does not belong.