RAN DISAGGREGATION · DEEP DIVE

From the monolithic LTE eNB to the disaggregated 5G CU-CP / CU-UP / DU / RU.

For 25 years one box did everything. Now four open software components — running on different hardware, sometimes from different vendors — speak F1, E1, and Open Fronthaul to deliver the same radio service. Here is every functional split, every interface, every reason the industry did this.

3GPP TS 38.300 · 38.401 · 38.473 · 38.463 · 38.425 O-RAN WG4 7.2x · WG3 E2 · WG2 A1 · WG1 O1 Specs 36.300 (LTE) · 23.214 (CUPS) · 37.340 (NSA)
PART 1 · WHAT WE HAD IN LTE

The monolithic eNB — one black box, one vendor.

In 4G LTE (3GPP Release 8 onward, 2008–present), the entire radio access network is delivered as a single physical unit called the eNodeB. It contains the radio frequency (RF) chain, the entire physical layer (PHY), every Layer-2 protocol (MAC, RLC, PDCP), the radio resource control (RRC) entity, and the connection to the core network — all wrapped in one vendor's hardware and one vendor's software. You buy eNBs from Ericsson, Nokia, Huawei, ZTE, Samsung, or NEC. You do not mix.

UE phone · CPE LTE-Uu RRC · PDCP · RLC · MAC · LTE-PHY eNB LTE BASE STATION · MONOLITHIC L3 RRC · S1AP · X2AP L2.4 PDCP — cipher · integrity · ROHC L2.3 RLC — AM · UM · TM · ARQ L2.2 MAC — HARQ · scheduler · BSR L1 LTE-PHY — full physical layer OFDMA · SC-FDMA · MIMO · turbo coding · RF · antenna S1-MME S1AP / SCTP / IP S1-U GTP-U / UDP / IP EPC EVOLVED PACKET CORE MME S-GW P-GW HSS
Figure 1 · LTE Release 8 monolithic architecture · UE ↔ eNB ↔ EPC · single S1 interface to control + data plane combined

What the eNB owned monolithic

  • RF + antenna — RRH (Remote Radio Head) or integrated
  • Entire PHY — OFDMA, MIMO, channel coding, channel estimation
  • MAC — HARQ, scheduling decisions, DRX, BSR/PHR
  • RLC — segmentation/reassembly, three modes (AM/UM/TM), ARQ
  • PDCP — header compression (ROHC), ciphering (SNOW3G/AES), integrity
  • RRC — connection state, mobility, measurement reporting
  • S1AP / X2AP — talking to MME and neighbor eNBs

What the eNB couldn't do limitation

  • Mix vendors — Ericsson RF + Nokia baseband? Not supported.
  • Independent scaling — to add coverage you bought a full eNB (RF + baseband)
  • Pool baseband — every cell needed its own processor; idle cells wasted CPU
  • AI/ML in the air interface — no open interface to inject decisions
  • Centralize signaling — RRC state was stuck at the cell site
  • Soft handovers across vendors — X2 worked only between identical-version eNBs
  • Cloud the baseband — proprietary hardware accelerators baked in
PART 2 · THE BRIDGE — CUPS

The first 3GPP split was actually in the core, not the RAN.

3GPP Release 14 (2017) introduced CUPS — Control and User Plane Separation for the EPC. The Serving Gateway became S-GW-C + S-GW-U; the PDN Gateway became P-GW-C + P-GW-U. A new control protocol called PFCP (on the Sxa/Sxb interfaces) let one S-GW-C steer many S-GW-Us. This was the proof-of-concept the industry used to justify full RAN disaggregation in 5G: scale control and data planes independently, on different hardware, ideally from different vendors.

UE eNBmonolithic(no change in CUPS) MME S-GW-Ccontrol P-GW-Ccontrol S-GW-Udata plane P-GW-Udata plane Internet LTE-Uu S1-MME S1-U S11 S5/S8-C Sxa PFCP Sxb PFCP S5/S8-U SGi
Figure 2 · CUPS (Rel-14) · gateways split into control (purple) and data plane (cyan) · PFCP on Sxa/Sxb · eNB itself still monolithic
PART 3 · WHAT WE HAVE IN 5G

The gNB becomes four things.

3GPP Release 15 (2018) shipped the full split. The 5G base station — the gNB — is logically not one node. It is up to four: CU-CP, CU-UP, DU, RU. Each one terminates a specific subset of the protocol stack, sits on its own hardware, and speaks well-defined 3GPP (and O-RAN) interfaces to its neighbors. Specs: TS 38.401 defines the overall architecture and the F1 + E1 interfaces; TS 38.473 is F1AP; TS 38.463 is E1AP; TS 38.425 is the F1-U user-plane container; O-RAN WG4 defines the Open Fronthaul 7.2x split that goes RU↔DU.

UEphone · IoT O-RUPHY-low · RFiFFT · beamforming · DAC O-DURLC · MAC · PHY-highscheduler · HARQ · LDPC O-CU-CPRRC · PDCP-CNGAP · F1AP · E1AP O-CU-UPSDAP · PDCP-UGTP-U · NR-U · ciphering AMFMobility (5GC) UPFData Plane (5GC) DNInternet Uu 7.2x Open FH F1-C F1AP/SCTP F1-U NR-U/GTP-U E1 E1AP/SCTP NG-C NGAP/SCTP NG-U GTP-U/UDP N6 IP — LOGICAL gNB (TS 38.401) — 5G NR — DISAGGREGATED gNB RU ↔ DU ↔ CU-CP / CU-UP · F1 · E1 · OPEN FH 7.2x · NG-C · NG-U
Figure 3 · 5G NR disaggregation · the gNB is now four logical functions over open interfaces · O-RAN names them O-RU / O-DU / O-CU-CP / O-CU-UP
PART 4 · WHERE DO YOU CUT THE STACK?

3GPP studied eight places to split. Two won.

TR 38.801 documented eight functional split candidates — each labeled Option 1 through Option 8 — describing where to draw the line between the central unit and the distributed unit. The trade-off is the eternal one in telecoms: centralization gain (cheaper, smarter, more poolable) vs fronthaul cost (the lower in the stack you split, the more bandwidth and the tighter the latency the link between them needs). The industry converged on Option 2 for the F1 interface (CU↔DU) and Option 7-2x for the Open Fronthaul (DU↔RU). The legacy Option 8 split is what CPRI did and what 5G is trying to escape from.

OPTION 1
RRC / PDCP
Highest split. Only RRC stays central; everything else distributed. Almost no centralization gain — never deployed.
unused
OPTION 2 ⭐
PDCP / RLC
The F1 split. CU has PDCP + RRC; DU has RLC + MAC + PHY. Most flexibility for fronthaul (gigabit-class IP transport, ~10 ms one-way budget). Industry standard for F1 today.
F1 · 3GPP TS 38.401
OPTION 3
High-RLC / Low-RLC
Splits RLC itself. Useful for some dual-connectivity (LTE+NR Bearer split) but rare as a CU/DU boundary.
niche
OPTION 4
RLC / MAC
Centralizes RLC, leaves MAC distributed. Trade-off doesn't pay — never widely adopted.
unused
OPTION 5
High-MAC / Low-MAC
Splits the scheduler. Theoretically allows centralized coordination scheduling, but latency budget < 1 ms — fragile.
research
OPTION 6
MAC / PHY
"Coordinated multipoint" candidate. Significant centralization gains, very tight FH latency <250 µs. Some C-RAN deployments use this.
C-RAN
OPTION 7-2x ⭐
PHY-low / PHY-high
DU keeps coding + modulation, RU does iFFT + beamforming + RF. The Open Fronthaul split. eCPRI over Ethernet. 10–25 GbE typical. Latency budget ~100 µs.
O-RAN WG4 · 7.2x
OPTION 8
PHY / RF
The legacy CPRI split. RU is a "dumb antenna" — DU does everything digital. ~25 Gbps per 100 MHz of bandwidth. Inflexible and expensive — what Open Fronthaul replaces.
CPRI · legacy
PART 5 · LTE VS 5G — THE FULL TABLE

What changed, line by line.

DimensionLTE (4G)5G NR (disaggregated)
Base station uniteNB (single physical box)gNB = CU-CP + CU-UP + DU + RU (4 logical functions)
Vendor modelVertical lock-in (one vendor end-to-end)Open multi-vendor (different vendors per unit)
Layer-3 (RRC)In eNBIn CU-CP
PDCP (cipher / integrity)In eNB (single instance for CP+UP)Split: PDCP-C in CU-CP · PDCP-U in CU-UP
SDAP (QoS-flow→DRB)Doesn't exist (LTE has no QFI)In CU-UP — new for 5G
RLC / MACIn eNBIn DU
PHY-high (coding, modulation)In eNBIn DU
PHY-low + RFIn eNB (or RRH over CPRI)In RU (Open Fronthaul 7.2x)
Control to coreS1-MME (S1AP / SCTP) — single interfaceNG-C / N2 (NGAP / SCTP) from CU-CP only
User-plane to coreS1-U (GTP-U / UDP)NG-U / N3 (GTP-U / UDP) from CU-UP only
Inter-site linkX2 (X2AP) for HO + dual connectivityXn (XnAP) — protocol-equivalent, different stack
Core networkEPC — MME / S-GW / P-GW / HSS / PCRF5GC — AMF / SMF / UPF / AUSF / UDM / PCF / NRF / ... (SBA)
Core-network interfaceDiameter (S6a) + GTP-C (S5/S11)HTTP/2 + JSON (SBI / SBA) — modern microservices
Control-plane split (CUPS)Added in Rel-14 for S-GW + P-GW onlyBuilt-in everywhere · SMF programs UPF via PFCP (N4)
Open RAN extensionsNone (closed by design)SMO · Non-RT RIC (A1) · Near-RT RIC (E2 + xApps)
Fronthaul transportCPRI (Option 8) · proprietaryeCPRI · Open FH 7.2x · IP/Ethernet · 25–100 GbE
Backhaul transportIP/Ethernet (microwave or fiber)IP/Ethernet — same, but stricter QoS / sync
Time-sync requirement≤±1.5 µs (PTP G.8275.1)≤±130 ns (TDD) · ≤±60 ns (URLLC) · LLS-C profiles
Deployment modelPhysical box per cell siteCloud-native (vRAN) · CU on common server · DU near antenna · RU at antenna
AI/ML integration pointNone definedRIC xApps (Near-RT) + rApps (Non-RT) via E2 + A1
PART 6 · EVERY NEW INTERFACE

The protocol stacks that didn't exist in LTE.

Disaggregation invented new wire formats — most live in 3GPP TS 38.4xx or O-RAN WG4. Here's exactly what's on each cable.

InterfaceBetweenL7 / EncapL4SpecWhy it exists
F1-CDU ↔ CU-CPF1APSCTP / IPTS 38.473Carries DU↔CU control: F1 Setup, UE Context, RRC msg transfer, paging
F1-UDU ↔ CU-UPNR-U inside GTP-UUDP / IPTS 38.425Carries PDCP PDUs DU↔CU with flow-control feedback
E1CU-CP ↔ CU-UPE1APSCTP / IPTS 38.463CU-CP tells CU-UP which bearers to set up, hands over PDCP keys + F1-U TEIDs
Open FH 7.2xRU ↔ DUeCPRI (CUS-plane + M-plane)Ethernet / IPO-RAN WG4 CUS / MCarries IQ samples + beamforming weights + sync + management
NG-C / N2CU-CP ↔ AMFNGAPSCTP / IPTS 38.413RAN ↔ Core control (NG Setup, Initial UE Msg, PDU Session Resource Setup, Handover)
NG-U / N3CU-UP ↔ UPFGTP-U with QFI ext headerUDP / IPTS 29.281RAN ↔ Core user-plane data with QoS-Flow marking
E2Near-RT RIC ↔ DU/CUE2AP + Service Models (KPM, RC, NI, CCC)SCTPO-RAN WG3Near-RT RIC consumes KPIs, sends RAN control commands at 10 ms–1 s timescale
A1Non-RT RIC ↔ Near-RT RICREST + JSONHTTPS / TCPO-RAN WG2 A1Pushes policies + ML models from SMO down to Near-RT RIC for xApps
O1SMO ↔ every NENETCONF + YANGSSH / TCPO-RAN WG1 O1FCAPS — fault, config, performance, security mgmt of every CU/DU/RU/NF
O2SMO ↔ O-CloudO-Cloud Mgmt APIHTTPSO-RAN WG6 O2Manages the underlying cloud (Kubernetes, OpenStack) that hosts vRAN workloads
PART 7 · WHY OPERATORS DID THIS

What disaggregation actually unlocks.

🏭

Multi-vendor RAN

RU from Mavenir, DU from Wind River, CU from Samsung, RIC from Juniper. Open interfaces mean operators can mix and match instead of buying one vendor end-to-end. Rakuten Mobile built an entire greenfield network this way.

Cloud-native baseband (vRAN)

DU and CU run as containers on COTS x86 (Intel FlexRAN) or Ampere ARM servers. Independent scaling: ten DUs sharing one CU; one CU pool covering hundreds of cells.

🤖

AI / ML at the air interface

Near-RT RIC hosts xApps over E2 — closed-loop control of scheduling, beamforming, anti-jamming, ML-based handover, energy saving. Non-RT RIC hosts rApps for slower policy + ML training. This was impossible in LTE.

Independent CP / UP scaling

Heavy data traffic? Scale CU-UP and UPF horizontally without touching CU-CP. Signaling storms? Scale CU-CP and AMF. Same gain CUPS gave the core, now in the RAN.

🔒

Per-domain security

Different keys, different operators, different trust zones per slice. CU-UP can be hosted at the edge for a hyperscaler tenant; CU-CP can stay in the operator core. Zero-trust between every interface.

🌐

Network slicing in the RAN

NSSAI flowing all the way to the DU scheduler. Different SLAs (eMBB, URLLC, mMTC) sharing one physical RAN with isolated logical instances per slice.

Want to build these architectures?

Drop these exact components onto a canvas and watch interfaces wire themselves. The Build Custom canvas in the simulator extras knows every 3GPP-defined pair from RU↔DU to AMF↔SEPP.

✏ Build Custom Topology → 📡 RAN Splits side-by-side 🎓 O-RAN Course (24 lessons)