From SISO PRB-Step Identification to MIMO 2×2 Bottleneck-Onset Research Design

The previously published Sci_O-RAN investigation characterised receiver-observed downlink throughput under six independent applied-PRB transitions in a virtual 5G/O-RAN Tb3 testbed. The present technical report develops a prospective two-input, two-output (MIMO 2 × 2) research design to explain performance saturation and identify changes in the limiting resource. Its two candidate interventions are the applied downlink PRB quota and independently controlled offered downlink traffic load; its proposed responses are application-layer downlink goodput and a semantically verified downlink queue-occupancy measure.This is system-identification MIMO, not a claim of two transmit and two receive radio antennas. The document specifies measurement boundaries, local linear models, non-linear saturation hypotheses, a layered telemetry-capability matrix, timestamp and sampling requirements, and competing explanations for bottleneck onset. Public srsRAN and Linux documentation identifies plausible observables but does not establish their implementation or collection readiness on the installed Tb3 environment. A staged experimental design separates capability assessment, minimal software adaptation, a bounded pilot, data acquisition, and independently held-out validation. The original SISO model selection and its limitations remain intact: no physical actuation delay, unique second-order plant dynamics, or statistically decisive superiority of the exploratory M3 model is asserted. No live testbed action, new data acquisition, or additional model fitting was undertaken for this document. The first repository-only telemetry inventory is explicitly deferred.