Why site-specific channel modeling matters once 6G radios start sensing
In sum – what to know:
- Two different questions – Standardized stochastic models benchmark a class of environments, while an RF digital twin predicts how a system behaves at one real site.
- ISAC raises the bar – Sensing turns multipath into data, adding detection, range, velocity, angle and tracking metrics that must stay physically consistent.
- Field trials get leaner – Validated twins and emulation let engineers test more scenarios earlier, leaving field measurements to confirm rather than discover.
For years, the wireless industry has leaned on standardized statistical models to test and validate radios in the lab. CDL (Clustered Delay Line) and TDL (Tapped Delay Line) models — essentially mathematical representations of how radio signals behave in different environments — have served as the backbone of 5G NR testing. The question now is whether they’re sufficient for 6G, particularly as integrated sensing and communication (ISAC) becomes a defining use case.
At the RCR Wireless 6G Forum, Keysight product manager Nelson Mayedo moderated a discussion with Usman Virk, an R&D manager and solution architect for RF Digital Twin Solutions, who laid out how the industry’s approach to channel modeling and emulation needs to evolve. Virk’s view is that existing methods remain valuable, but 6G radios that sense their surroundings, use massive antenna arrays, coordinate across technologies, and make AI-driven decisions will demand more than average statistical performance. Site-specific channel modeling, real-world measurements, and hardware-based emulation all need to be part of the picture.
The takeaway from the session was fairly clear. Confidence in 6G won’t come from any single validation method. It’ll come from a continuous workflow that combines standardized models, RF digital twins, field measurements, and channel emulation into one coherent pipeline. That’s a more complex proposition than what 4G or 5G required, and it’s worth understanding why.
Statistical models still matter, but 6G needs site-specific testing
Standardized stochastic models, the kind the industry has relied on for years, represent a class of propagation environments. They define channel characteristics statistically, making them well suited for benchmarking, Monte Carlo evaluation, and reproducible comparisons across technologies and implementations. They aren’t tied to a specific location, and that generality is both their strength and their limitation.
Site-specific RF digital twins, such as ray tracing, take the opposite approach. They model a particular physical environment using actual geometry, materials, antenna configurations, and user mobility. Rather than telling you how a system performs on average across a category of environments, a digital twin tells you how it’s expected to perform in a specific deployment.
Virk offered a useful analogy during the session. “A standardized stochastic model is something like a test track, while RF digital twin is a virtual representation of the actual environment where the system will eventually operate,” he said. He compared the same idea to autonomous vehicle testing — a generic traffic model is useful, but eventually you have to test specific roads, obstructions, and moving objects. For 6G, where radios may need to understand performance behind a specific obstruction or track a moving sensing target, that distinction matters. Average statistical performance doesn’t always cut it.
That said, more detail isn’t automatically better. The required fidelity should match the engineering question and the evaluation KPI. A full ray-traced digital twin is overkill for some validation tasks, and standardized models remain the right tool for system-level comparisons. Virk was explicit on this point. “The future is not stochastic versus deterministic modeling competition,” he said. “It’s about selecting or combining the models when it is appropriate to best match the validation objective.” The two approaches answer different questions, and neither replaces the other.
ISAC changes what channel emulation must reproduce
ISAC fundamentally changes what the propagation environment means for a radio. In conventional communications, the channel is something signals pass through on the way from transmitter to receiver — engineers evaluate how reliably information transfers from one point to another. In ISAC, the received multipath also carries information about objects, their motion, and the surrounding environment. The echoes themselves become data.
That distinction has real implications for what channel models need to capture. A realistic ISAC scenario must combine sensing paths through targets, background multipath, and clutter alongside the full communication channel. It may also require radar cross section, delay, direction, Doppler, micro-Doppler, antenna characteristics, and the dynamics of moving targets. Compared to 4G and 5G link validation — where the focus is on throughput, BLER (block error rate), and SINR (signal-to-interference-plus-noise ratio) — ISAC adds an entirely new dimension of sensing metrics such as detection probability, range accuracy, velocity estimation, angle accuracy, and tracking performance.
Generating all of these effects is one challenge. Maintaining their physical and temporal consistency as the transmitter, receiver, targets, and environment all move relative to each other is a harder one. Virk drew an analogy to illustrate the point. “You can think of it like having a conversation in a large hall, while using the echoes to locate a moving object,” he said. “Both the conversation and the echoes must be represented accurately.”
The consequences of getting this wrong are worth flagging. If the channel model isn’t physically consistent, sensing algorithms can appear to perform exceptionally well in testing — unrealistically well. An overly clean or simplified channel can make detection and tracking look easy, masking problems that would show up in a real deployment. That’s a counterpoint worth keeping in mind as the industry builds out ISAC test frameworks. Channels that are too idealized make a successful test result mean very little.
This multidimensional validation requirement is what makes ISAC channel emulation fundamentally harder than conventional link emulation. You’re no longer validating a single set of communication KPIs. You’re simultaneously validating communication performance and sensing accuracy, and the two are coupled through the same physical channel.
Digital twins and emulation complement field trials
Once a site-specific RF digital twin has been built and a scenario defined, the resulting channel data can feed into two distinct validation workflows. Software-in-the-loop lets engineers evaluate algorithms, architectures, and system behavior before complete hardware is available. Hardware-in-the-loop introduces real devices — base stations, chipsets, sensing receivers — while a channel emulator recreates controlled, repeatable RF propagation conditions around that hardware.
The continuity across those workflows is the key advantage. An engineer can define a scenario, create an RF digital twin, reproduce the same channel conditions through RF emulation, test real hardware, and benchmark against field measurements — all within a single pipeline. Software-in-the-loop acts as a virtual rehearsal; hardware-in-the-loop brings real equipment onto the same virtual stage. The channel emulator ties them together by recreating the radio environment around the hardware in a way that’s repeatable and controllable.
Field trials, of course, remain essential. No simulation fully replaces reality. But field testing is expensive, difficult to repeat exactly, and constrained by weather, access, hardware availability, and environmental conditions. Reproducing the exact same target movement, blockage, or interference scenario in the field is practically impossible. Validated digital twins and emulation workflows let engineers test more scenarios earlier, reproduce failures on demand, and compare competing implementations under identical channel conditions. Field measurements then play a more strategic role — parameterizing and validating the model, establishing confidence, and verifying final system behavior rather than repeatedly surfacing problems late in development.
Virk compared the approach to wind tunnel testing in aviation. “It doesn’t eliminate the flight testing completely. It makes flight testing more focused, repeatable, and informative,” he said. “RF digital twins and channel emulation can play exactly a similar role.”
There’s a historical argument worth considering here, too. Mayedo floated it himself during the session, saying some people might argue 4G took eight or ten years to become commercially available because test methods weren’t quite optimal. Whether that timeline could have been meaningfully compressed is debatable, but the broader point holds. Earlier, more systematic validation — enabled by digital twins and emulation — can support faster design decisions, continuous regression testing, and controlled comparisons of competing approaches. For 6G, where the validation surface is substantially larger than previous generations, that kind of efficiency matters more than it did before. Whether it’s a prerequisite for hitting commercial timelines without cutting corners on reliability is the open question.
Table of Contents
