L
Admin
[Technical article, rfforum Editorial Team]
A channel emulator will happily pass a UE or CPE that later fails on a highway or in a dense indoor cell. The usual cause is a delay-spread / Doppler profile that is too kind, or frozen while you only stressed MCS and power.
1. Delay spread is not "more taps = harder"
What matters is the power-delay profile shape and the excess delay relative to your equaliser / cyclic prefix. A long, low-power tail can be easier than a short profile with two strong, closely spaced paths that the equaliser fights. Match the profile to the deployment class you claim (urban macro, indoor hotspot), not to whatever default the rack shipped with.
2. Doppler and time variation
Static or very slow fading hides tracking-loop and AGC problems. If the product will move (vehicular, drone, handheld walk), set a Doppler consistent with carrier frequency and speed, and let the channel run long enough that AGC, timing, and beam/CSI tracking actually adapt.
3. Correlation and MIMO
Independent fading per antenna is often optimistic. Spatial correlation and cross-polar ratios change rank and EVM under MIMO. If the datasheet quotes 2x2 or 4x4 rates, the emulator correlation matrix should be part of the recorded test condition.
4. Report the profile
Write the model name (or tap powers/delays), Doppler, correlation, and run duration next to throughput / BLER. Otherwise a "pass" is not reproducible.
Discussion: which standard channel model (or custom PDP) do you treat as the minimum bar for FR1 outdoor-ish claims, and do you ever re-run with a harsher Doppler after a clean lab pass?
— rfforum 编辑部 / Editorial Team
A channel emulator will happily pass a UE or CPE that later fails on a highway or in a dense indoor cell. The usual cause is a delay-spread / Doppler profile that is too kind, or frozen while you only stressed MCS and power.
1. Delay spread is not "more taps = harder"
What matters is the power-delay profile shape and the excess delay relative to your equaliser / cyclic prefix. A long, low-power tail can be easier than a short profile with two strong, closely spaced paths that the equaliser fights. Match the profile to the deployment class you claim (urban macro, indoor hotspot), not to whatever default the rack shipped with.
2. Doppler and time variation
Static or very slow fading hides tracking-loop and AGC problems. If the product will move (vehicular, drone, handheld walk), set a Doppler consistent with carrier frequency and speed, and let the channel run long enough that AGC, timing, and beam/CSI tracking actually adapt.
3. Correlation and MIMO
Independent fading per antenna is often optimistic. Spatial correlation and cross-polar ratios change rank and EVM under MIMO. If the datasheet quotes 2x2 or 4x4 rates, the emulator correlation matrix should be part of the recorded test condition.
4. Report the profile
Write the model name (or tap powers/delays), Doppler, correlation, and run duration next to throughput / BLER. Otherwise a "pass" is not reproducible.
Discussion: which standard channel model (or custom PDP) do you treat as the minimum bar for FR1 outdoor-ish claims, and do you ever re-run with a harsher Doppler after a clean lab pass?
— rfforum 编辑部 / Editorial Team
