QoE-based Traffic Steering Across Mobile and Wi-Fi Networks

Since it’s Halloween time, let’s share what tends to scare operators the most with Wi-Fi Offloading.

The biggest concern is the “Device is King” issue. Operators often ask:

• What if the device connects to Wi-Fi even when cellular connection is good or better?
• If both mobile and Wi-Fi have strong quality, the device might prefer Wi-Fi, leaving the mobile network underutilized?

Both concerns about user experience and control are valid, but somewhat exaggerated. Devices are increasingly smarter at selecting the best connection—Wi-Fi or cellular. Over the years, we’ve observed this trend; however, it is indeed true that devices frequently auto-connect to known Wi-Fi networks—authenticated via the SIM card—when both cellular and Wi-Fi networks provide good quality connections.

To address concerns about losing control and preserving user experience, 3GPP has specified several architectures designed to grant more control—all of which involve backhauling traffic to the mobile core. Ultimately, however, the decision of network selection still rests with the device.

The only 3GPP standard that has seen widespread deployment is untrusted non-3GPP access, primarily used for Wi-Fi Calling. Very few operators utilize the trusted non-3GPP access architecture for backhauling data traffic to the mobile core. Instead, around 90% of Enea’s MNO and MVNO customers prefer to implement local traffic break-out. One reason is that the 3GPP-specified backhauling architecture is overly complex, requiring significant investments in specialized gateway functions such as WAG, TWAG, or TNGF. Additionally, backhauling typically only offloads the Radio Access Network (RAN), leaving the mobile core unaffected (And, many times more loaded with traffic due to better Wi-Fi coverage).

Move Contol to the Edge with Enea Aptilo SMP

With the unique capabilities of the Enea Aptilo SMP, we have long helped our customers overcome control issues by shifting management to the network edge rather than relying on backhauling traffic:

• We can retrieve policies from PCRF/PCF and enforce them locally via Wi-Fi gateways.
• When charging for Wi-Fi services, operators can manage quotas at the edge, with SMP checking out time and data quotas from BSS charging systems, including options for top-ups or returning unused quotas.
• Traffic can be redirected to central deep packet inspection (DPI) nodes or, when necessary, to lawful enforcement nodes, this is however not backhauling in the 3GPP sense.
• Many DPI use cases can also be handled at the edge, as modern Wi-Fi gateways from leading vendors often feature built-in DPI capabilities, some enabled by Enea’s iXEngine embedded DPI technology.

Solving “The Device Is King” Issue

To address the “device is king” challenge and shift network selection control to the mobile operator, 3GPP has defined the Access Network Traffic Steering, Switching, and Splitting (ATSSS). Often regarded as the “Holy Grail” of offloading, ATSSS envisions using both Wi-Fi and mobile networks as a unified data plane, giving operators greater control from the mobile core. However, while promising, its implementation may never materialize—largely due to high costs and the ongoing political battle between MNOs and device manufacturers over who should be in control. The risk is that ATSSS could follow a similar path as the Access Network Discovery and Selection Function (ANDSF), remaining a theoretical concept rather than a practical solution.

That is precisely what our innovation—QoE-based Traffic Steering Across Mobile and Wi-Fi Networks—delivers.

Nothing changes—the device will still decide whether to connect to Wi-Fi or mobile networks. However, our solution ensures that the device is not the sole decision-maker in network selection; the final decision is with the operator. We call this approach “Intelligent Rightloading,” and it comes in two variants.

Predefined Traffic Steering

• Allows operators to prevent Wi-Fi connections for specific user groups, locations, times of day, or based on policies and historical data, including AI-based congestion predictions.
• Data Sources: Imported historical data, integration with analytics nodes such as NWDAF in 5G SA, and AI-powered congestion prediction functions within traffic management systems like the Enea Traffic Management solution for mobile networks.
• Use Cases:
  A. Prevent offloading during times and/or in locations where the mobile network is not congested.
  B. Restrict offloading to certain subscription types.
  C. Load balancing—e.g., only offload every third user.

Real-time QoE-based Traffic Steering

• Cellular-centric: Offload traffic to Wi-Fi only when the mobile session quality falls below a defined threshold.
  • Data sources (mobile):
    1: Traffic management systems, such as Enea TM, providing session-specific real-time quality data from the mobile network.
    2: Future CAMARA API (work in progress).
    4: Probe-based measurements.
• Holistic approach: Continuously collect real-time quality data from both cellular and Wi-Fi networks. Offload traffic to Wi-Fi only if it offers better quality or if load balancing is required.
  • Data sources for mobile as above.
  • Data sources (Wi-Fi):
    3a: Proprietary integrations with specific Wi-Fi vendors.
    3b: Upcoming standards—WBA is working on a harmonized method for delivering quality KPIs over RADIUS.
    4. Probe-based measurements.