LTE
4G
3Factors: RRC · S1SIG · E-RAB
Fam 1RRC_ESTAB (RRC setup)
Fam 27S1SIG (logical S1)
Fam 8·9ERAB_ESTAB / _ADD
14E-RAB failure causes
6EstabCause dimensions
The Samsung accessibility KPI — ErabAccessibilityInit (verbatim)
From the Counter Description KPI chapter: "This KPI shows the probability for an end-user to be provided with an E-RAB at request." Samsung computes it — per cell, every measurement period — as the product of the three stage success-rates, plus a companion failure-rate KPI:
KPI family "KPI · Accessibility" — exact Samsung formulas
| KPI (Samsung name) | Unit | Formula (verbatim, granularity) |
|---|---|---|
| ErabAccessibilityInit | % | (SumRrcConnEstabSucc ÷ SumRrcConnEstabAtt) × (SumS1sigS1ConnEstabSucc ÷ SumS1sigS1ConnEstabAtt) × (SumErabEstabInitSuccNbr ÷ SumErabEstabInitAttNbr) × 100 |
| ErabAccessibilityAdd | % | ΣQCI SAEB.NbrSuccEstabAdd.QCI ÷ ΣQCI SAEB.NbrAttEstabAdd.QCI × 100 (added E-RABs on an existing call) |
| ErabConnectionFailureRate | % | 100 − ErabAccessibilityInit (with zero-guards: all three denominators 0 → 0; any one 0 → 100) |
≡
Why a product — the three factors are sequential gates. 99.9 % × 99.9 % × 97.2 % = 97.0 %: the aggregate can only be as good as the weakest stage, and the decomposition tells you which family's failure counters to open next. That is the entire method of this page.
The same KPIs on the operator board (verbatim from the LTE KPI reference workbook)
Reporting KPIs — exact formulas as run in daily reporting
| KPI (as reported) | Formula (verbatim) |
|---|---|
| Session Setup Success Rate (%) | [(ConnEstabSucc/ConnEstabAtt) × (S1ConnEstabSucc/S1ConnEstabAtt) × (EstabInitSuccNbr/EstabInitAttNbr)] × 100 — the same three-factor product at raw-counter level |
| RRC Connection Success Rate (%) | [ConnEstabSucc ÷ ConnEstabAtt] × 100 |
| S1 Establish Success Rate (%) | [S1ConnEstabSucc ÷ S1ConnEstabAtt] × 100 |
| ERAB Established Success Rate (%) — All QCI | [(EstabInitSuccNbr + EstabAddSuccNbr) ÷ (EstabInitAttNbr + EstabAddAttNbr)] × 100 |
| RRC Setup Failure Rate (%) | ConnEstabFail_RrcSigTo ÷ SumRrcConnEstabAtt × 100 — the failure KPI is keyed directly to the air-interface timeout counter |
| RRC Setup Completion Rate (%) | RrcEstabMsg_RrcConnectionSetupComplete ÷ RrcEstabMsg_RrcConnectionSetup × 100 — from message-counter Family 905 RRC_ESTAB_MSG |
| RRC Re-establishment Success Rate (%) | RrcRrcConnectionReestablishmentComplete ÷ RrcRrcConnectionReestablishmentRequest × 100 |
Where each counter pegs — the collection points on the call flow
Samsung's Counter Description defines every counter by its collection time on the call flow. The three attempt/success pairs peg at these exact messages:
Attempt/Success peg points — RRC → S1SIG → E-RABper Counter Description figures
attempt/success collection pointsRRC & E-RAB are indexed by EstabCause; E-RAB additionally by QCI — so every SR can be split per cause and per service.
The EstabCause index (verbatim) — every RRC/E-RAB counter splits on it
| Value | Cause | Meaning (verbatim) |
|---|---|---|
| 0 | emergency | Emergency call connection |
| 1 | highPriorityAccess | Access by the UE with a specific access class |
| 2 | mt_Access | Access by the UE due to paging received |
| 3 | mo_Signaling | Access by the UE due to attach |
| 4 | mo_Data | Access by the UE due to service request |
| 5 | delay_tolerant_access | Access by the delay tolerant access |
Read the counter name — the failure-domain decoder
Samsung failure counters are named <Stage>Fail_<Domain><Cause>. The prefix after the underscore identifies the failure domain — learn the four prefixes and any failure counter on this page (or in LSM) tells you where to look before you open a single log:
Prefix → failure domain → first place to look
| Prefix | Domain | What it means | First place to look |
|---|---|---|---|
Cp… | Control plane (call control) | ECCB/ECMB call-control logic: timeouts (CpCcTo), block reset / cell disabled (CpCcFail), CAC rejections (CpCapaCacFail, CpQosCacFail, CpBhCacFail) | Cell operational state (RTRV-CELL-STS), CAC settings, eNB SW restarts |
Up… | User plane (protocol blocks) | A protocol block returned Fail or failed audit: UpMacFail, UpRlcFail, UpPdcpFail, UpGtpFail | Channel-card / DSP health, transport (GTP), board alarms |
RrcSig… | Air interface (UE ↔ eNB) | The UE never answered: RrcSigTo (SetupComplete / ReconfigurationComplete timeout), RrcSigFail (re-establishment mid-procedure) | RF coverage, UL interference, PRACH/PUCCH health — the radio problems |
S1ap… | S1 / core side | S1 link or MME: S1apLinkFail (SCTP OOS / SCTB reset), S1apSigTo (MME silent), S1apSigFail (S1AP Reset received), S1apCuFail (S1AP-spec cause), S1apMmeOvld | SCTP/S1 transport, MME health & dimensioning |
!
This is the optimization shortcut — a dip in any stage SR is never investigated as a whole. Rank that stage's failure counters, take the top one, and its prefix routes you:
RrcSig… → drive-test/RF; Up… → hardware/transport; Cp… → capacity/config; S1ap… → core. Different prefixes = different teams. KPI ① — RRC Connection Establishment (Family ID 1 · RRC_ESTAB)
The KPI. First factor of ErabAccessibilityInit: SumRrcConnEstabSucc ÷ SumRrcConnEstabAtt × 100. The family counts "access statistics between RRC connection request and RRC connection setup completions, collected for each cell", split by EstabCause. The procedure (per feature LTE-SW0320, RRC Connection Management): RRCConnectionRequest arrives on SRB0 / CCCH in TM mode; the eNB allocates SRB1 and answers RRCConnectionSetup; the UE completes on SRB1 / DCCH in AM mode, carrying the NAS Attach/Service Request that the eNB forwards as the Initial UE Message.
①-a · Attempt & success counters (verbatim descriptions)
| Counter | Pegs when (verbatim trigger) |
|---|---|
| ConnEstabAtt | "cumulated when the eNB received the RRCConnectionRequest from the UE" |
| ConnEstabSucc | "cumulated when the eNB received the RRCConnectionSetupComplete" |
i
Companion families — Family 905 RRC_ESTAB_MSG counts the raw messages (
RrcEstabMsg_RrcConnectionRequest / RrcConnectionSetup / RrcConnectionSetupComplete) — Setup-vs-Complete gives the pure air-interface completion rate. Family 5 RRC_CONN (ConnNoAvg/Max/Tot/Cnt) holds the connected-UE load you'll correlate CAC rejects against. Recovery lives in Family 3 RRC_REESTAB (ConnReEstabAtt/Succ + the same Fail/Reject cause split, plus InterEnb… variants with rejects like RlfHoNotPossible / ErrIndication). ①-b · Failure counters — call released after RRCConnectionSetup was sent
| Counter | Domain | Exact cause (verbatim from the collection-time figures) |
|---|---|---|
| ConnEstabFail_CpCcTo | Cp | Call-control timeout — no response from the protocol blocks (MAC, RLC, PDCP, GTP) |
| ConnEstabFail_CpCcFail | Cp | ECMB reset · cell operational state (RTRV-CELL-STS) → Disabled · UE-info audit mismatch between ECCB and RLC/MAC · available MME doesn't exist |
| ConnEstabFail_UpMacFail | Up | MAC returned Fail, or ECCB↔MAC audit mismatch on the UE |
| ConnEstabFail_UpRlcFail | Up | RLC returned Fail, or audit mismatch with the RLC/DSP |
| ConnEstabFail_UpPdcpFail | Up | PDCP returned Fail after RRCConnectionSetup was transmitted |
| ConnEstabFail_RrcSigTo | RrcSig | Timer expiry — RRCConnectionSetupComplete never received from the UE |
| ConnEstabFail_S1apLinkFail | S1ap | Call-related S1AP link went Out of Service, or SCTB reset, mid-procedure |
| ConnEstabFail_S1apSigFail | S1ap | S1AP RESET received from the MME mid-procedure |
①-c · Reject counters — eNB sent RRCConnectionReject instead of Setup
| Counter | Exact cause (verbatim) |
|---|---|
| ConnEstabReject_CpCcFail | Cell Disabled (RTRV-CELL-STS) · internal resource allocation failure · invalid MME |
| ConnEstabReject_CpCapaCacFail | CAC based on Call Count per eNB or Call Count per Cell rejected the call (feature LTE-SW4101) |
| ConnEstabReject_S1apMmeOvld | UE-Identity is S-TMSI and its MME is in OverLoad — or UE-Identity is random and all MMEs are OverLoad (feature LTE-SW0503) |
①-d · Optimization — the RRC-stage playbook
Top failure counter → technique (features & MOs are Samsung-verbatim)
| Dominant counter | Diagnosis | Technique / lever |
|---|---|---|
ConnEstabFail_RrcSigTo | UE unreachable after Setup — UL coverage / interference, not capacity | RF first: UL SINR / RSSI, PRACH & PUCCH health, drive test. Check per-EstabCause: if only mt_Access suffers, suspect paging coverage |
ConnEstabReject_CpCapaCacFail | Call-count CAC exhausted | Feature LTE-SW4101: MO [ENB] enb-cac-info-func / cell-cac-info-func — max-enb-call-count, max-call-count, call-cac-threshold-for-normal (emergency/handover keep their own thresholds). Raise within HW limits or offload/split the cell |
ConnEstabReject_S1apMmeOvld | MME told the eNB to shed load | Feature LTE-SW0503 MME Overload Control: eNB obeys S1AP Overload Start Action IE per EstablishmentCause and rejects with waitTime back-off. Fix = MME capacity / pooling — not an eNB parameter |
ConnEstabFail_S1apLinkFail / SigFail | S1/SCTP instability | Check SCTP association, transport flaps, SCTB resets; correlate with S1SIG family (below) — the same event usually pegs both |
ConnEstabFail_Up*Fail | Protocol-block (board) trouble | Channel-card / DSP audit failures — check board alarms & SW restarts; hardware ticket, not parameter tuning |
ConnEstabFail_CpCcFail spikes | Cell disabled events / block restarts | Correlate with RTRV-CELL-STS history and ECMB resets — availability problem masquerading as accessibility |
i
Overload-protection denies are counted separately — when eNB overload protection (not the MME) throttles access, the denied/discarded requests peg Family 253 DENIED_CALL:
Denied_HighPriorityAccess / Denied_moSignaling / Denied_moData / Denied_moVoiceCall / Denied_DelayTolerantAccess / Denied_Paging / Denied_EmergencyCall (+ the matching Discarded_* set), gated by overloadProtectCtrl, thresholdForNormalCall and the per-cause protectPerEstablishCause thresholds. If users complain but RRC SR looks clean, read these. KPI ② — UE-associated logical S1 Connection Establishment (Family ID 27 · S1SIG)
The KPI. Second factor: SumS1sigS1ConnEstabSucc ÷ SumS1sigS1ConnEstabAtt × 100. After RRC completes, the eNB forwards the UE's NAS message to the MME as the Initial UE Message — the attempt. Success pegs when the MME answers with any of three messages (all three collection-time figures are in the Counter Description): Initial Context Setup Request, Downlink NAS Transport, or UE Context Release Command. This factor isolates "is the MME answering the eNB at all?" from radio and bearer problems.
②-a · Attempt & success counters (verbatim descriptions)
| Counter | Description (verbatim) |
|---|---|
| S1ConnEstabAtt | "UE logical S1 connection setup attempt count" — pegs at Initial UE Message |
| S1ConnEstabSucc | "UE Logical S1 Connection Setup success count" — pegs at ICS Request / DL NAS Transport / UE Ctx Release Cmd |
②-b · Failure counters (verbatim descriptions)
| Counter | Domain | Exact cause (verbatim) |
|---|---|---|
| S1ConnEstabFail_S1apSigTo | S1ap | "released due to S1AP signaling timeout (not received) during S1 connection" — the MME never answered the Initial UE Message |
| S1ConnEstabFail_S1apLinkFail | S1ap | "released due to S1 SCTP link failure during S1 connection" |
| S1ConnEstabFail_S1apSigFail | S1ap | "released due to receiving S1AP signaling during S1 connection" (e.g. Reset) |
| S1ConnEstabFail_S1apCuFail | S1ap | "released due to the S1AP specification cause during S1 connection" |
| S1ConnEstabFail_CpCcFail | Cp | "released due to reset notification (eNB failure or block restart) from ECMB or by the ECCB block during S1 connection" |
②-c · Optimization — the S1-stage playbook
Top failure counter → technique
| Dominant counter | Diagnosis | Technique / lever |
|---|---|---|
S1apSigTo | MME slow or silent — MME CPU, paging storm, or transport latency | MME-side capacity & dimensioning; verify with MME logs; if one MME of the pool, check S-TMSI distribution / MME weight factors |
S1apLinkFail | SCTP association drops | Transport QoS on the S1 path (backhaul flaps, firewall SCTP timeouts); SCTP heartbeat/RTO settings; redundancy (multi-homing) |
S1apSigFail / S1apCuFail | MME actively resetting / spec-cause releases | Decode the S1AP cause in traces — TAC misconfiguration, UE context conflicts, MME SW issues; coordinate with core team |
CpCcFail | eNB internal reset during S1 setup | ECMB/ECCB restart history — same treatment as RRC ①: availability, not tuning |
!
Signature to memorise — RRC SR healthy + S1SIG SR degraded +
S1apSigTo dominant = pure core/transport problem. No RF or eNB parameter will move it. This decomposition is why Samsung multiplies three factors instead of publishing one CSSR blob. KPI ③ — E-RAB setup (Family ID 8 · ERAB_ESTAB / Family 9 · ERAB_ESTAB_ADD, per cell · EstabCause · QCI)
The KPI. Third factor: SumErabEstabInitSuccNbr ÷ SumErabEstabInitAttNbr × 100. The family covers "the default bearer setup to activate the initial DRB's radio and S1 bearer". Per the Counter Description: Attempt pegs when the Initial Context Setup Request is received; Success when the Initial Context Setup Response is transmitted; "a failure is counted when a timeout or failure occurs during the procedure and the applicable cause value is recorded." In between sit the UE-capability exchange, SecurityModeCommand/Complete and RRCConnectionReconfiguration/Complete — each with its own failure counters. A second sub-family, E-RAB Setup Add (EstabAddAttNbr/SuccNbr, pegged on E-RAB Setup Request), covers bearers added to an existing call and feeds ErabAccessibilityAdd.
The E-RAB setup window — everything between ICS Request and ICS Responsepeg points ③
failure anywhere in the window pegs an
ErabInitFailNbr_* causeAdded bearers: same window from E-RAB Setup Request → ErabAddFailNbr_*. ③-a · Attempt & success counters (verbatim)
| Counter | Sub-family | Pegs when |
|---|---|---|
| EstabInitAttNbr | E-RAB Setup (initial) | Initial Context Setup Request received (per cell · EstabCause · QCI) |
| EstabInitSuccNbr | E-RAB Setup (initial) | Initial Context Setup Response transmitted |
| EstabAddAttNbr | E-RAB Setup Add | "Success/Failure of E-RAB after receiving ErabSetupRequest message" — attempt |
| EstabAddSuccNbr | E-RAB Setup Add | "Success of E-RAB after receiving ErabSetupRequest message" |
i
Companion families — Family 10 ERAB_TIME (
EstabTimeAvg/Max/Tot/Cnt) times the setup; Family 833 ERAB_INFO_PLMN_QCI repeats the whole Init/Add/Mod counter set per PLMN and QCI (ErabInfo_EstabInitAttNbr… + modification failures ErabInfo_ModFailNbr_*) — that's your per-operator/per-service accessibility view on MOCN sites. The Add family also has one extra cause the Init set doesn't: ErabAddFailNbr_CpCcInteraction. ③-b · The 14 failure causes — ErabInitFailNbr_* (ErabAddFailNbr_* mirrors them)
| Counter | Domain | Exact trigger (verbatim from the collection-time figures) |
|---|---|---|
| ErabInitFailNbr_CpCcTo | Cp | Call-control timeout during the setup window |
| ErabInitFailNbr_CpCcFail | Cp | e.g. CSFB to a not-supported RAT type in the UE context setup |
| ErabInitFailNbr_CpCapaCacFail | Cp | Rejected by Capacity CAC (LTE-SW4101) |
| ErabInitFailNbr_CpQosCacFail | Cp | Rejected by QoS CAC (LTE-SW4102 — GBR admission on PRB usage) |
| ErabInitFailNbr_CpBhCacFail | Cp | Rejected by Backhaul CAC (backhaul bandwidth budget exhausted) |
| ErabInitFailNbr_UpGtpFail | Up | GTP layer setup fail — S1-U tunnel could not be created |
| ErabInitFailNbr_UpMacFail | Up | MAC layer setup fail |
| ErabInitFailNbr_UpRlcFail | Up | RLC layer setup fail — or ARQ ack not received on traffic |
| ErabInitFailNbr_UpPdcpFail | Up | PDCP layer setup fail |
| ErabInitFailNbr_RrcSigTo | RrcSig | Timeout waiting for SecurityModeComplete or RRCConnectionReconfigurationComplete |
| ErabInitFailNbr_RrcSigFail | RrcSig | UE went to re-establishment in the middle of the reconfiguration |
| ErabInitFailNbr_S1apCuFail | S1ap | Invalid E-RAB information in the ICS Request |
| ErabInitFailNbr_S1apLinkFail | S1ap | S1 went Out of Service mid-setup |
| ErabInitFailNbr_S1apSigFail | S1ap | Security-config conflict (Integrity == EIA0 && Ciphering != EEA0) — or UE Context Release Command / S1AP Reset received mid-setup |
③-c · Optimization — the E-RAB-stage playbook (the CAC triad)
Top failure counter → technique (features verbatim from SVR24B Feature Description)
| Dominant counter | Diagnosis | Technique / lever |
|---|---|---|
CpCapaCacFail | UE/bearer-count ceiling | LTE-SW4101 — decision is current UEs < MaxUeCELL (max-call-count) / MaxUeENB (max-enb-call-count) scaled by call-cac-threshold-for-normal (emergency & handover use their own thresholds, so they survive congestion). Raise limits per channel-card capability, or offload |
CpQosCacFail | GBR admission (PRB budget) | LTE-SW4102 — GBR bearers admitted against PRB usage; operator caps PRB / backhaul BW / GBR-bearer count. Tune the GBR budget, or use LTE-SW4103 Preemption + LTE-SW4106 CAC per QCI to protect priority services |
CpBhCacFail | Backhaul bandwidth budget | Backhaul CAC rejecting — verify configured backhaul bandwidth vs actual link; increase transport or adjust the budget |
UpGtpFail | S1-U tunnel setup failing | Transport to S-GW: routing, MTU, GTP-U path; correlate with S-GW availability — this is the classic "RRC fine, no data bearer" signature |
RrcSigTo / RrcSigFail | UE lost during security/reconfig | Radio again — cell-edge UEs failing at the reconfiguration step; check coverage & UL quality; per-QCI split shows if VoLTE (QCI-1) is hit |
S1apSigFail (EIA0 case) | Security algorithm mismatch | Config: integrity EIA0 with non-EEA0 ciphering is rejected by design — align eNB/MME security algorithm configuration |
CpCcFail (CSFB case) | CSFB to unsupported RAT | Check CSFB target RAT config vs what the ICS Request demands — feature/neighbour configuration, not radio |
≡
Read ③ per QCI and per cause — the family is indexed
cNum · EstabCause · QCI. ErabAccessibilityInit dropping only on QCI 1 = VoLTE admission/GBR problem (SW4102 budget); only on mt_Access = paging-triggered calls failing late (coverage); across all QCIs with CpCapaCacFail = raw capacity. The same headline number, three different fixes. Inside the eNB — the CPS blocks the counters are talking about
Every failure cause above names an internal software block. Per the LTE eNB System Description (Ver. 9.0): the Call Processing Software (CPS) consists of the eNB Control processing Subsystem (ECS) — "responsible for network access and call control" — and the eNB Data processing Subsystem (EDS) — "responsible for user traffic handling". ECS = SCTB (SCTP), ECMB (common management), ECCB (call control), CSAB (SON agent), TrM (trace/CSL). EDS = GTPB, PDCB, RLCB, MACB.
CPS structure — and which counter prefix each block feedsSystem Description Fig. 38
the failure counters are literally block-fail reports
ConnEstabFail_CpCcFail's "ECMB reset / ECCB audit" and UpMacFail's "Fail message from the MAC" name these exact blocks.≡
Admission control, verbatim — the System Description defines both CAC types the counters reject on: capacity-based ("a threshold for the maximum number of connected UEs (new calls/handover calls) and bearers") →
…CpCapaCacFail, and QoS-based ("estimated PRB usage of the newly requested bearer … and the maximum acceptable limit of the PRB (per bearer type, QCI, and UL/DL)") → …CpQosCacFail. ECCB runs CAC; ECMB runs overload control & access barring (Family 253's Denied_*). The master matrix — the failure-counter routing fabric
Every failure counter is one wire from a domain (who owns the fault) to a stage (where it pegs). Click a domain to light its blast-radius across all three stages, a stage to see everything that fails there, or a scenario to watch a real-world signature ripple across the families.
Failure-Counter Routing Fabric
Scenarios
The admission & barring gate pipeline — every access-control feature (SVR24B)
A UE's access request passes four gates in order — barring → overload protection → CAC → RRC procedure. Each node below is a real Samsung feature; click any node to open its full card: what it is, the benefit, the recommended optimization settings, its key MOs and the exact counters it moves. When accessibility drops, walk the pipeline left-to-right and rule out the deliberate gates before tuning capacity.
Access-Control Gate Pipelineclick a feature node to open its detail card
!
Order matters — a request passes barring → overload protection → CAC → procedure. When SR drops, first rule out the deliberate gates (barring active? SIB14? overload action from MME?) before tuning capacity: rejected-by-design is not a fault.
Accessibility Optimization Studio — counters → grade → root cause → Samsung fix
Enter the raw counters for a cell; the studio computes the Session Setup Success Rate (the three-factor product), grades every stage against Samsung's alert thresholds, finds the weakest link, and prints the exact parameter changes to make — path, current → recommended, expected gain — all taken verbatim from the Samsung LTE Accessibility Optimization guide. Nothing here is invented: the levers are the real MOs, the gains are Samsung's own estimates.
Accessibility Optimization Studiolive
① RRC · Family 1
② S1SIG · Family 27
③ E-RAB · Family 8
Session Setup Success Rate
—%
RRC × S1SIG × E-RAB
Recommended Samsung actions
!
Order of work — fix in the order the studio ranks it, and change one lever at a time: re-measure the target counter after each move. Samsung's own rollout sequences these over three weeks (see the TOP-10 table's Implementation Week column) — CAC switches first, RACH power second, timers third.
The optimization levers — TOP-10 accessibility parameters (verbatim)
Samsung's ranked accessibility parameters, exactly as published — current vs recommended, the expected gain, and the rollout week. These are the levers the studio recommends.
TOP-10 accessibility parameters — rank · current → recommended · gain
| # | Parameter | Current → Recommended | Gain | Week |
|---|---|---|---|---|
| 1 | drb-count-cac-enable | disable(0) → enable(1) | +3–5% | 1 |
| 2 | drb-cac-threshold-for-normal | 90.00% → 85.00% | +2–4% | 1 |
| 3 | preamble-initial-received-target-power | −120 dBm → −104 dBm | +4–6% | 2 |
| 4 | qos-cac-option | qos-cac-no-use(0) → qos-cac-use(1) | +2–3% | 1 |
| 5 | max-drb-count (QCI 1) | 400 → 120 | +3–4% | 3 |
| 6 | power-ramping-step | 4 dB → 2 dB | +2–3% | 2 |
| 7 | call-count-cac-enable | use(1) → use(1) ✓ (verify on) | +2–3% | 1 |
| 8 | ra-response-window-size | sf10 → sf7 | +1–2% | 2 |
| 9 | rrc-connection-setup (timer) | 19000 ms → 15000 ms | +1–3% | 3 |
| 10 | zero-correlation-zone-config | 1 → 5 | +2–4% | 2 |
Detailed configuration — range · default · counter impact (verbatim)
| Parameter | Range | Default | Recommended | Counter impact |
|---|---|---|---|---|
| drb-count-cac-enable | disable(0) | enable(1) | disable(0) | enable(1) | ⬇ ErabInitFailNbr_CpCapaCacFail by 60–80% |
| drb-cac-threshold-for-normal | 0.00 … 100.00 | 90.00% | 85.00% | ⬇ ErabInitFailNbr_CpCapaCacFail by 50–70% |
| preamble-initial-received-target-power | −120 … −90 dBm | −120 dBm | −104 dBm | ⬆ RACH SR 89 → 95% |
| qos-cac-option | no-use(0) | use(1) | no-use(0) | use(1) | ⬇ ErabInitFailNbr_CpQosCacFail by 60–80% |
| max-drb-count | 0 … 600 | 400 | QCI1:120 · QCI5:80 · QCI8:200 | ⬇ service-specific failures 40–60% |
| power-ramping-step | 0 … 4 dB | 4 dB | 2 dB | ⬆ RACH attempt efficiency by 25% |
| ra-response-window-size | sf2 … sf10 | sf10 | sf7 | ⬇ access delay by 30% |
| rrc-connection-setup | 10 … 65535 ms | 19000 ms | 15000 ms | ⬇ RRC timeout failures by 20% |
| zero-correlation-zone-config | 0 … 15 | 1 | 5 | ⬇ preamble collisions by 60% |
| prach-config-index | 0 … 63 | 0 | 3 | ⬆ PRACH density → success rate |
i
Hierarchy — RACH levers live under
eutran-cell-fdd-tdd/rach-config-common; E-RAB CAC under …/cell-admission-control/cell-plmn-cac-info-func and …/qci-drb-cac-info; eNB-level call CAC under enb-function/admission-control/enb-cac-info-func; the RRC setup timer under enb-function/emtc-timer-info. Root-cause library — why each stage fails (verbatim)
RACH poor success — the four sub-procedures (Msg1→Msg4)
| Sub-procedure | Root causes (Samsung, verbatim) |
|---|---|
| Preamble (Msg1) not detected | Poor UL RF (low UE Tx power at cell edge, high UL path-loss, UL interference); incorrect PRACH config (sparse prach-ConfigIndex, bad rootSequenceIndex/zeroCorrelationZoneConfig, preambleInitialReceivedTargetPower too low); preamble collisions (too few contention preambles, high load); timing errors |
| RAR (Msg2) not received | Poor DL RF (low PDCCH/PDSCH power, DL interference); ra-ResponseWindowSize misconfig; UE fails to decode RA-RNTI PDCCH |
| Msg3 Tx/Rx fail | Poor UL RF on the PUSCH carrying Msg3; UL grant too small / wrong power; HARQ decode fail after retransmissions; mac-ContentionResolutionTimer expiry |
| Contention (Msg4) fail | Poor DL RF on Msg4; UE can't decode its identity; eNB scheduler can't grant Msg3 fast enough under PRACH overload |
RRC poor success — Samsung root causes (verbatim)
| Failure point | Root causes |
|---|---|
UE can't get RRCConnectionSetup | Poor DL RF on PDCCH/PDSCH; UE can't apply the config (incompatible params / capability / bug); T300 expiry at the UE |
eNB can't get SetupComplete | Poor UL RF on PUSCH; eNB decode failure; eNB monitoring-timer expiry (pegs ConnEstabFail_RrcSigTo) |
| Admission rejects before Setup | Max RRC connections reached; core-network rejection of the InitialUEMessage (not subscribed, auth fail, MME overload); eNB congestion/overload |
Monitoring & alert thresholds (verbatim)
Target and Critical/Major/Minor alarm bands — the studio grades against these
| Metric | Counter formula | Target | Minor | Major | Critical |
|---|---|---|---|---|---|
| E-RAB Service Accessibility | (EstabInitSucc+EstabAddSucc)/(EstabInitAtt+EstabAddAtt)×100 | ≥ 98% | < 99% | < 98% | < 95% |
| RRC Setup Success Rate | ConnEstabSucc / ConnEstabAtt ×100 | ≥ 98% | < 99% | < 98% | < 95% |
| RACH Success Rate | Random Access success | ≥ 95% | < 97% | < 95% | < 90% |
| Capacity CAC Failure Rate | ErabInitFailNbr_CpCapaCacFail / EstabInitAttNbr ×100 | ≤ 2% | > 2% | > 3% | > 5% |
| QoS CAC Failure Rate | ErabInitFailNbr_CpQosCacFail / EstabInitAttNbr ×100 | ≤ 1% | > 1% | > 2% | > 3% |
Worked case study — RRC Setup failures from signalling timeouts (verbatim)
Case: RRC Setup SR 92% ←
ConnEstabFail_RrcSigTo 6%Samsung guideProblem
RRC Setup SR at 92% vs 98% target; RRC Setup Failure Rate 6%, dominated by ConnEstabFail_RrcSigTo; RRC Setup Time elevated.Hypothesis
Suboptimal RRC timers + inefficient SRB RLC config → messages not exchanged in time → rrc-connection-setup timer expiry.Outcome
RRC Setup SR +2–3%; lower ConnEstabFail_RrcSigTo and reduced average RRC setup time. the fix set
rrc-connection-setup 19000→15000 ms · timer-poll-retransmit ms200→ms150 · max-retransmission-threshold t32→t16 · timer-reordering ms200→ms160 (SRB RLC) Optimization feature trials — real Samsung feature IDs
Features that move the accessibility KPIs
| Stage | Feature ID | Name · purpose (verbatim) |
|---|---|---|
| RACH | LTE-SW4105 | Access Class Barring — manages UE access to PRACH by access class under overload |
| RACH | LTE-SW2102 | eNB Overload Control (Adaptive Access Barring) — dynamic ACB to control RA load |
| RACH | LTE-SW0315 | Extended Access Barring (SIB14) — selectively bar MTC/NB-IoT to cut RA collisions |
| RACH | LTE-SO0401 | RACH Optimization — SON tuning of preambles, backoff, response timers |
| RACH | LTE-ME0603 | Random Access Procedures — CBRA/CFRA mechanism & parameter enhancements |
| RRC | LTE-SW0326 | Connection Setup Time Enhancement — parallel Security Mode Command + RRC Reconfiguration |
| RRC | LTE-SW0322 | E-RAB Management — coordinates bearer setup with RRC, cuts setup time & failures |
| RRC | LTE-SW0320 | RRC Connection Management — incl. RRE control for CPU-overload protection under signalling storms |
| RRC | LTE-SW0316 | Network Signaling (NS) Support — reduces overload failures from excessive signalling |
One KPI, three gates — decompose, decode the prefix, fix the domain.
That's the whole LTE accessibility method: ErabAccessibilityInit → weakest factor → that family's top failure counter → prefix routes the fix. Next, see how the same thinking maps to the NR leg in NSA and the full 5GC chain in SA.
Sources: Samsung eNB Counter Description for SLR 4.5 v7.0 & LTE eNB Counter Description for SLR 6.0.0 v3.0 (counter names, triggers, KPI formulas, collection-time figures) · LTE Samsung KPI.xlsx — operator KPI reference workbook whose "eNB Counters" sheet is the LTE eNB Counter Description in table form (Family IDs 1 / 27 / 8 / 9 / 905 / 3 / 5 / 10 / 833 / 253 and reporting formulas verbatim) · LTE eNB System Description Ver. 9.0 (Airtel) (CPS/ECS/EDS block architecture, ECCB/ECMB functions, admission-control definitions) · eNB (LTE) Feature Description for SVR24B (LTE-SW0320/0503/0508/4101–4106, CAC MOs & parameters). The SVR25A Counter Description xlsx and SVR25A Troubleshooting Manual are NASCA-DRM-protected and could not be read; counter families shown are verified across the readable releases and the operator workbook.