By Claus Hetting, WiFi NOW CEO & Chairman
Today’s wireless performance is not constrained by compute power but by legacy RF designs – in particular for FEMs – that attempt to treat symptoms instead of rethinking the technology itself. RF inefficiency can no longer be corrected digitally and ‘after the fact’ without triggering thermal runaway. The state-of-the-art solution is intelligent, self-contained analog RF architectures that are millions of times faster than any digital correction. These are the key messages in a recent paper authored by QuantalRF.
What’s holding back Wi-Fi performance more than anything else? According to RF solutions pioneer QuantalRF, the answer is legacy RF designs. The advent of AI is accelerating the need for analog RF innovation because AI workloads are setting new and much more rigorous standards for latency and data bandwidth. Performance levels required by AI can no longer be reached with current RF designs that use digital corrections to patch up inefficiencies. And the industry can no longer afford to kick the can down the road, says Dr Ali Fard, CEO & CTO of QuantalRF in a new paper.
“For years, wireless systems have leaned on digital fixes – pre-distortion, calibration, interference mitigation – combined with material tweaks and brute-force power to mask RF inefficiency. These techniques worked only because system margins were wide enough to tolerate delay, inefficiency, and excess power. Digital could afford to be slower than the electromagnetic world it was correcting. That approach bought time at the cost of rising heat, power, and complexity. That tradeoff is now breaking,” Dr Ali Fard says.
The fundamental issue is that digital systems by their nature are ‘blind’ to instantaneous RF behaviour and therefore cannot interact with the RF waveform itself. “By the time the system “sees” the problem, the RF waveform has already moved on. As AI workloads push systems toward tighter latency budgets, higher duty cycles, and higher sustained throughput, the delay between RF disturbance and digital correction becomes fatal rather than tolerable,” he says.
QuantalRF says that the industry didn’t ignore these fundamental RF issues but instead postponed trying to find a solution – mostly because RF technology is extremely difficult to master. “Every major player wants the same result: Lower power, lower cost, higher performance, and more reliable wireless links. Almost none move decisively to achieve it. Unsurprisingly, most organisations retreat to safer ground: Incremental tweaks instead of architectural change, more digital compensation instead of better RF, brute-force margins, and material workarounds. Those approaches buy time, but they don’t scale. They avoid architectural risk – and in doing so, lock in architectural stagnation,” Dr Ali Fard says.
So what is the solution? QuantalRF says big industry players have spent billions in an attempt to push RF (FEM) designs into silicon (CMOS), a change that would introduce dramatically lower versus incumbent GaAs and SiGe. But despite massive engineering efforts, legacy front-end designs proved unstable in silicon, Dr Ali Fard says. Instead QuantalRF solutions are based on regenerative and super-regenerative RF architectures, systems that were originally researched and designed for NASA by Dr Forrest Brown. The breakthrough came in 2021 when QuantalRF’s Mats Carlsson – today VP of Engineering – and Dr Ali Fard created multi-feedback RF architecture.
“QuantalRF could only exist because the mathematics, simulation capability, and system-level understanding finally caught up with the physics. Ideas once dismissed as too unstable, too complex, or too fragile could now be controlled, scaled, and manufactured in CMOS,” Dr Ali Fard says.
For more make sure to download the full paper below by filling out the form and don’t miss Dr Ali Fard speaking at Wi-Fi World Congress in Mount View, California, on April 14.
