Traffic Steering in 5G and MEC Solutions: End-to-End Workflow

One of the biggest promises of 5G is its ability to deliver application-aware connectivity, where traffic is dynamically steered toward the optimal path, slice, or computing resource. This becomes especially critical in Multi-access Edge Computing (MEC) scenarios, where latency-sensitive applications like AR/VR, autonomous vehicles, or industrial automation require traffic to be offloaded to the edge rather than going to a centralized data center.

Here’s how traffic steering is achieved end-to-end, across the RAN, Transmission Network (TN), and Core Network (CN).

1. RAN Level Steering

At the Radio Access Network (RAN), traffic steering begins with QoS Flow management:

• Each user session is associated with a QoS Flow Identifier (QFI), which is mapped to the corresponding Data Radio Bearer (DRB).

• The gNB applies Radio Resource Management (RRM) to enforce slice-specific QoS, ensuring that, for example, URLLC traffic gets prioritized scheduling and minimal delay.

• RAN slicing allows different flows (eMBB, URLLC, mMTC) to be assigned to dedicated logical slices.

• For MEC, the gNB can classify uplink traffic based on uplink classifier (UL CL) rules, sending it to a local UPF close to the edge rather than the centralized core.

Key RAN Enablers:

• 5QI-based QoS mapping

• UL CL for edge breakout

• RAN slicing policies (configured by RAN Intelligent Controller or O-RAN SMO)

2. Transmission Network (TN) Steering

The Transmission Network (fronthaul, midhaul, backhaul) ensures that traffic steering decisions from the RAN are carried forward with proper isolation and QoS.

• Segment Routing (SR-MPLS or SRv6) provides path steering by assigning traffic to specific label stacks or segments.

• HQoS (Hierarchical QoS) ensures that slices and flows get bandwidth guarantees and hierarchical priority across transport links.

• eFlex Ethernet slicing allows flexible bandwidth partitioning per slice, mapping RAN slice traffic to transport slices.

• DetNet (Deterministic Networking) may be applied for ultra-reliable low-latency services requiring strict jitter control.

Key TN Enablers:

• SRv6/MPLS for slice-based routing

• HQoS for intra-slice prioritization

• Ethernet slicing for bandwidth isolation

3. Core Network (CN) Steering

In the 5G Core, traffic steering is primarily handled by the Session Management Function (SMF) and User Plane Function (UPF):

• SMF applies policies from the Policy Control Function (PCF) to determine if traffic should be anchored in a central UPF or steered to a local UPF near a MEC.

• UL CL and Branching Point UPF: Traffic can be directed based on the destination IP/application to the appropriate UPF. For example, video streaming might be anchored at a central UPF, while AR/VR flows are offloaded at an edge UPF.

• PCF and NEF (Network Exposure Function): Allow dynamic traffic rules based on application requests or enterprise API triggers.

Key CN Enablers:

• SMF/UPF with CUPS (Control/User Plane Separation)

• UL CL for MEC traffic offloading

• PCF for dynamic policy control

• NEF for third-party application-driven steering

4. MEC Application Layer

Once traffic is steered to the MEC:

• MEC hosts local applications (AR/VR server, factory controller, video optimizer).

• The Local UPF provides direct breakout to the MEC data network, bypassing central data centers.

• MEC orchestrator coordinates with 5GC to ensure applications scale with user demand and remain aligned with slice SLAs.

Conclusion

Traffic steering in 5G + MEC is an end-to-end process:

• RAN classifies and prioritizes traffic via QoS flows and uplink classifiers.

• TN enforces slice isolation and SLA through SRv6, HQoS, and Ethernet slicing.

• Core applies SMF/UPF anchoring rules to decide whether traffic stays central or offloads at the edge.

• MEC executes application workloads closer to the user, ensuring ultra-low latency and high performance.