Network & Telecom

Control Plane

Q: What is the difference between the control plane and the user plane?

The control plane carries signaling that establishes, modifies, and tears down sessions and computes forwarding rules, while the user plane carries the actual subscriber payload (voice frames, IP packets, video) according to those rules. In 5G, control-plane messages such as NAS registration, RRC connection setup, and PFCP session establishment travel over the N1, N2, and N4 reference points, whereas user data rides GTP-U tunnels on N3. Control-plane traffic is low volume but latency- and reliability-sensitive: a single RRC reconfiguration may be only a few hundred bytes, yet it must complete within milliseconds to avoid call drops. User-plane traffic is high volume, often gigabits per second per gNB, but tolerates more buffering. This separation, formalized as Control and User Plane Separation (CUPS) in 3GPP Release 14, lets operators scale the two planes independently.

Q: How does control-plane latency affect handover and call setup?

Control-plane latency is the time to transition a device from idle to a connected, data-ready state, and 3GPP sets a target of under 10 ms for 5G NR versus roughly 50 ms in LTE. This budget covers RRC connection setup, security activation, and bearer establishment. For mobility, the X2 or Xn handover signaling exchange between source and target base stations typically completes in 20 to 50 ms; if control-plane messages are delayed by congestion or poor radio conditions, the user-plane interruption time grows and TCP throughput collapses. Ultra-reliable low-latency (URLLC) services therefore prioritize control-plane resources and may pre-provision context at candidate cells (conditional handover) so the signaling exchange is shortened at the moment of cell change.

Q: What protocols run in the control plane of a 5G network?

The radio control plane uses RRC (Radio Resource Control) between the UE and gNB, riding on PDCP, RLC, and MAC layers. Above the radio, NAS (Non-Access Stratum) messages flow between the UE and the AMF for registration and mobility. On the core side, the Service Based Architecture uses HTTP/2 with JSON over the SBI between functions such as AMF, SMF, and PCF, while the SMF programs the UPF user plane through the PFCP protocol over N4. In an SDN deployment, OpenFlow or P4Runtime carries control-plane instructions from a centralized controller down to switching hardware. All of these are signaling protocols: they move intelligence and state, not bearer payload.

/kuhn-TROHL playn/

Signaling, routing intelligence, and session state all live in this functional layer of a communications network, which computes the forwarding rules that the data-carrying user plane then applies to subscriber traffic. In a 5G radio access network it runs RRC between the device and the gNB and relays device-to-AMF NAS signaling; in the core it covers the AMF, SMF, and PCF; and in software-defined networking it is abstracted into a centralized controller. Separating control from bearer traffic, formalized as Control and User Plane Separation in 3GPP Release 14, lets operators scale signaling and throughput independently and place each function where latency and capacity demand. Control-plane messages are low volume (hundreds of bytes per transaction) but reliability- and latency-sensitive, with 5G targeting under 10 ms for idle-to-connected transition.

Category: Network & Telecom

Latency target (5G): < 10 ms

Signaling protocols: RRC / NAS

How the Control Plane Steers a Network

Every packet-switched network divides its work into two cooperating layers. The control plane is the decision-making layer: it runs the protocols that discover topology, authenticate subscribers, negotiate quality-of-service, set up and tear down sessions, and compute the forwarding tables. The user plane is the execution layer that moves payload according to those tables. The control plane carries almost no traffic volume by comparison, yet it determines whether the user plane works correctly at all. In a cellular network, a single dropped or delayed control message can prevent a call from connecting or cause a handover to fail, so control-plane reliability targets are far stricter than raw throughput.

In 3GPP 5G architecture the radio control plane uses Radio Resource Control (RRC) between the device and the gNB, carried over PDCP, RLC, and MAC, while Non-Access Stratum (NAS) signaling flows between the device and the Access and Mobility Management Function (AMF). The core network adopts a Service Based Architecture in which control functions such as the AMF, Session Management Function (SMF), and Policy Control Function (PCF) communicate over HTTP/2 with JSON payloads. The SMF then programs the User Plane Function (UPF) using the Packet Forwarding Control Protocol (PFCP) over the N4 reference point, the clearest example of the control plane writing rules into the user plane.

Software-defined networking pushes this idea further by centralizing control logic in a controller that has a global view of the network and programs simpler forwarding hardware through OpenFlow or P4Runtime. The same separation appears in IP routing, where protocols such as OSPF and BGP form the control plane that builds the routing table, while line-card ASICs do the per-packet lookup in the user plane. Whether in radio, core, or transport, the pattern is identical: intelligence concentrates in the control plane, and forwarding speed concentrates in the user plane.

Control-Plane Latency Budget

Latency in the control plane is measured as the time to bring a device from idle to a connected, data-ready state, and as the signaling exchange time during mobility events. These budgets directly bound how quickly a session starts and how briefly a handover interrupts the user plane.

Control-plane (idle-to-connected) latency:
T_cp = T_RACH + T_RRC setup + T_security + T_bearer

5G NR target: T_cp < 10 ms (LTE ≈ 50 ms)

Handover interruption (user plane):
T_HO ≈ T_signaling + T_sync + T_RA

Signaling load per session:
L_cp ≈ N_msg × S_msg, where S_msg ≈ 100 to 600 bytes

T_RACH = random-access exchange, T_RA = target-cell random access, N_msg = number of signaling messages, S_msg = message size. Example: a conditional handover trims T_signaling by pre-provisioning UE context at the target cell.

Control Plane Versus User Plane

Attribute	Control Plane	User Plane
Carries	Signaling, routing, session state	Subscriber payload (voice, IP, video)
5G protocols	RRC (radio) / NAS (core)	SDAP (radio) / GTP-U (transport)
5G core function	AMF, SMF, PCF	UPF
Reference points	N1, N2, N4 (PFCP)	N3, N6 (GTP-U)
Traffic volume	Low (hundreds of bytes/txn)	High (Gbit/s per gNB)
Key metric	Setup latency < 10 ms, reliability	Throughput, packet delay budget
Scales with	Number of sessions / signaling rate	Aggregate data rate

Common Questions

Frequently Asked Questions

What is the difference between the control plane and the user plane?

The control plane carries signaling that sets up, modifies, and tears down sessions and computes forwarding rules; the user plane carries the actual payload according to those rules. In 5G, control messages (NAS, RRC, PFCP) travel over N1, N2, and N4, while user data rides GTP-U on N3. Control traffic is low volume but latency- and reliability-critical; user traffic is high volume (Gbit/s) but tolerates buffering. CUPS, defined in 3GPP Release 14, lets operators scale the two planes independently.

How does control-plane latency affect handover and call setup?

Control-plane latency is the idle-to-connected transition time, targeted under 10 ms in 5G NR versus roughly 50 ms in LTE, covering RRC setup, security activation, and bearer establishment. For mobility, X2/Xn handover signaling completes in about 20 to 50 ms; if it stalls, user-plane interruption grows and TCP throughput drops. URLLC services prioritize control-plane resources and use conditional handover, pre-provisioning context at candidate cells to shorten the exchange at cell change.

What protocols run in the control plane of a 5G network?

The radio control plane uses RRC between the device and gNB over PDCP/RLC/MAC, with NAS between the device and AMF for registration and mobility. The core Service Based Architecture uses HTTP/2 with JSON over the SBI among AMF, SMF, and PCF, and the SMF programs the UPF user plane via PFCP over N4. In SDN deployments, OpenFlow or P4Runtime carries control instructions from a centralized controller to forwarding hardware. All of these move state and intelligence, not bearer payload.

Related Terms

← Control Limit Control Signaling →