Measure the channel
Use available radio observations such as RSRP, RSRQ, SINR and CSI/CQI, subject to testbed access and interface availability.
THE DECISION CHAIN
An estimate is only the beginning.
KashefTel’s proposed pipeline follows the effect of radio intelligence all the way to communication performance.
Estimate useful state
Explore AI-based estimation where direct observations are imperfect. Compare with a defined baseline under the same conditions.
Inform adaptation
Evaluate how the estimate influences rank adaptation and beamforming. Track the resulting MCS, BLER and HARQ behavior.
WHY THE FULL PIPELINE MATTERSLower estimation error is useful.
Lower estimation error is useful.
Better link behavior is the goal.
Evaluation must include the cost of inference, stale information, changing channels and integration constraints. These tradeoffs define whether an approach is useful in a real system.
FROM SIGNAL TO EXPERIENCE
Measure the consequences.
Estimation quality → adaptation decisions → BLER / HARQ → useful throughput / latency. Each connection needs evidence.