Guide 14
Capacity diagnostics
Use per-cell ROP KPIs and prerequisite checks to form testable hypotheses.
Prepare an OSS KPI export
Capacity Diagnostics requires current Max or Enterprise authorization. Check the KPI CSV before selecting it; selection starts an authorized processing request. Review warnings and thresholds after results appear. A Max run spends three shared credits; authorized Enterprise/admin/facilitator processing is unlimited.
Provide cell_id, a parseable timestamp with timezone and meaningful ROP-level KPI columns. Numeric fields contain numbers only; PRB/PDCCH are percentages and throughput is Mbps. Avoid quoted delimiters inside fields: this parser uses simple comma/semicolon/tab separation.
Interface walkthrough
Start a capacity review
Use per-cell ROP KPI data, not a drive-test route CSV. Results are threshold-based hypotheses to verify.
cell_id,timestamp,rrc_connected_avg,prb_util_dl_pct,pdcch_blocking_pct,timing_advance_avg,user_throughput_dl_mbps,vswr,pim_flag,hw_alarm_flag,sleeping_cell_flag,plmn_mismatch_flag
EXAMPLE_A,2026-10-01T08:00:00+07:00,20,65,1,12,8,0,0,0,0,0Hygiene before classification
Review VSWR, PIM, hardware alarms, sleeping-cell and PLMN-mismatch evidence before diagnosing. Hygiene fields are clean/flag inputs in this importer: 0/false/no/empty mean clean; any other marker is flagged. Do not enter raw numeric VSWR ratios as if they were directly thresholded here.
Missing hygiene columns trigger a warning. Missing primary numeric KPI values can default to zero in the current parser, so complete the relevant KPI fields before interpreting the output. One synthetic row only demonstrates format; it does not meet the default 100-samples-per-cell requirement.
Tune, classify and verify
Set network-specific TA, throughput, PRB, PDCCH-blocking and minimum-sample thresholds. Inspect averages, hygiene flags, sample counts and reasons for every cell before accepting a class. Retuning the diagnosis is a review of the same input, not a new import.
The classes are coverage-limited, capacity-limited, scheduler-limited, healthy, excluded-hygiene and insufficient-data. They are screening hypotheses, not confirmed root causes. Export CSV to preserve the evidence and repeat the investigation with OSS counters and field measurements.