Course/Day 1/Foundations & Optimization Workflow

Day One

Foundations & Optimization Workflow

Build the shared baseline — Samsung hardware, the NR/LTE architecture, the managed-object parameter model, and the KPI-driven optimization method used all week.

◎

Day objective — the Samsung RAN portfolio, the architecture and parameter model, and how counters → KPIs drive the whole optimization loop. Pick a sub-topic on the left — it opens here on its own.

1.1

Samsung RAN Hardware Portfolio

The radio and baseband hardware you optimize — from Massive MIMO down to femto small cells — and the 3GPP band map that anchors every cell you touch.

Samsung site hardware — radio → baseband → transport → corewire view

data / fronthaul CPRI/eCPRI splitSmall cells attach to the same core as the macro layer.

Pick a sub-topic — the ones with › open their own detailed deep-dive page:

Radios — MMUs / RRUs + interactive port explorer (every connector per radio) · RF Tx/Rx chain · SYS/OPT/ANT/RET LED decoder · the real AU MO tree (e-tilt · power · RSSI alarms — SVR25B)
Baseband — CDU plus the CU/DU split hardware + the PTP/SyncE clock-unit MOs
Small cells — CB1000 (Link Cell, indoor), LTM (indoor mmWave · n257/258/261), FRMM (outdoor)
3GPP band map — 5G: n78 · n41 · n77 · n28; LTE: B3 · B7 · B40

3GPP band map — the exact frequencies behind the course bands

These are the standard 3GPP band definitions for the bands the course targets — the duplex mode alone already dictates a different optimization posture (TDD needs slot-format / interference discipline; FDD does not).

Band map (3GPP TS 38.101 / 36.101)

Band	RAT	Duplex	Frequency range	Common name
`n78`	5G NR	TDD	3300–3800 MHz	C-band (mid-band capacity)
`n41`	5G NR	TDD	2496–2690 MHz	2.5 GHz
`n77`	5G NR	TDD	3300–4200 MHz	C-band (extended)
`n28`	5G NR	FDD	703–748 / 758–803 MHz	700 MHz (coverage)
`B3`	LTE	FDD	1710–1785 / 1805–1880 MHz	1800 MHz
`B7`	LTE	FDD	2500–2570 / 2620–2690 MHz	2600 MHz
`B40`	LTE	TDD	2300–2400 MHz	2300 MHz

≡

Why hardware first — the MMU's beamforming (SSB beam sweep, RI/CQI behaviour), the TDD bands' slot format, and the small-cell/macro layering all set hard limits on what parameter tuning can achieve later in the week. You optimize within the hardware envelope.

1.2

Architecture & Parameter Model

How Samsung structures the network in software — the 5G NR SA gNB split and its interfaces, the LTE eNB recap, the managed-object hierarchy, and how parameters, FGR features and counters relate.

5G NR SA gNB — CU-CP / CU-UP / DU / RU split with F1 · E1 · NG3GPP TS 38.401

F1 / E1 intra-gNB (DU↔CU, CU-CP↔CU-UP) NG gNB↔5GC (N2 control · N3 user)

The split matters for optimization because each function owns different parameters and counters. RRC timers and admission live in the CU-CP; the scheduler, HARQ and RACH live in the DU; bearer/QoS handling lives in the CU-UP. When a KPI moves, the split tells you which node's managed object to open.

Managed-Object hierarchy — real Samsung MO path (from the Parameter Description)

Every parameter is a leaf at the end of a managed-object path. This is the actual path to the RRC timer t300 in the Samsung DU model — you navigate exactly this tree in LSM:

managed-element / gnb-du-function / gutran-du-cell / gutran-du-cell-entries /
ue-timer-constants / t300 = t-300-ms200 (default)

How parameter ⇄ feature (FGR) ⇄ counter ⇄ KPI relate

The optimization data model — what you change, what you measureSamsung model

5G NR SA — gNB with CU-CP / CU-UP / DU split; F1, E1, NG interfaces
LTE eNB — architecture and S1 / X2 recap
Managed-Object hierarchy — managed-element / gnb-du-function / gutran-du-…
The relationship — how parameters, features (FGR) and counters relate to one another
S1 & MME pool — the live EMS "S1" tab decoded + the term-point-to-mme config · SCTP · S1-flex pooling · states & troubleshooting

1.3

KPI Optimization Methodology

The method used all week: the five KPI families, Samsung counter families & Family IDs, the optimization loop, and reading PM exports into a KPI dashboard.

★

The full method, animated — the single-variable trial loop as a fully-animated React-Flow cycle, the answer-verification loop, the KPI-family classifier, Samsung conventions and an interactive self-audit checklist.Open the Optimization Method ›

Samsung Optimization Method — animated loop · verification · KPI families · self-audit
KPI Master — 280 KPIs, interactive — every KPI with real num÷denom PEGs + Family IDs · filter by RAT/domain · search any counter · per-PLMN MOCN + slice layers

The optimization loop — every trial follows this cyclebaseline → roll back

The five KPI families & their generic Samsung formula

KPI families

Family	Question it answers	Direction	Representative KPI (Family ID)
Accessibility	Can the UE get on?	↑	RRC Conn Est SR (`5213`)
Retainability	Does it stay connected?	↓ drop	Call-Drop / RLF Rate
Mobility	Does handover work?	↑	HO Success Rate
Integrity	Is the link healthy?	↓ BLER	Latency / BLER / VoNR
Throughput	How fast is it?	↑	DL/UL Throughput

Success-rate KPI = ( success counter ÷ attempt counter ) × 100 — e.g. RRC Est SR = RRC Setup Complete ÷ RRC Setup Attempt

The 5 KPI families — Accessibility, Retainability, Mobility, Integrity, Capacity (+ the Per-PLMN & Slice cross-layers), each with its Family IDs
Samsung counters — how the counter docs work: Category→Family→Index→Type, the four counter types (pmCC/pmSI/pmGAUGE/pmOW), the naming decoder
Optimization loop — how a KPI is built (event→PEG→ROP→formula) + the L0→L4 tune/verify loop & the four calc rules
Dashboards / explorer — the interactive 280-KPI dashboard: filter by RAT/domain, search any counter, read every formula
Samsung tools — Trace & CSL — the OAM stack (EMS/USM/LSM/CIMA/TCE), Call Trace, Call Summary Log, OMRT/SON/CognitiV & which tool for which job

Live KPI success-rate calculatorcomputes on your counters

KPI (Samsung Family ID)

Attempt counter

Success counter

SR = success ÷ attempt × 100

Success rate

98.60

RRC Connection Establishment SR

Family 5213

Samsung success-rate KPIs all follow this form — the selector relabels the KPI & Family ID; the arithmetic runs on the counter values you enter. Exact per-release counter PEG names come from the SVR-specific Counter Description.

LAB

Hands-on Lab 1 — Parameter Tree & KPI Computation

Navigate the Samsung parameter tree, locate a feature's parameters, pull the counters and compute three sample KPIs by hand — the muscle memory for the rest of the week.

Hands-on Lab 1

Navigate the parameter tree · pull counters · compute 3 KPIs

⚡

1Navigate the managed-object / parameter tree and locate a feature's parameters
2Pull the relevant counters from a PM export
3Compute three sample KPIs from their counter families and Family IDs
4Sanity-check the results against the KPI dashboard

LSMPM exportParameter tree

Deliverable — three computed KPIs traced from counters, ready to baseline the rest of the course.