Wi-Fi Offloading, How? – Chapter 4.1
Local Break-Out: The Dominant Deployment Model
Historically, not all standardized functions—like those defined in 3GPP—make it into real-world networks. Vendors and service providers only implement features when there are strong commercial incentives. The integration of Wi-Fi with 3G and 4G data planes is a prime example. Most mobile operators have implemented local break-out (LBO) for Wi-Fi traffic directly from their secure Wi-Fi networks (802.1x).
Offloading the entire cellular network, rather than just the RAN, is preferable to backhauling traffic to the mobile core. Backhauling entails complex architectures and requires additional specialized gateways, as well as increased capacity costs in existing packet gateways.
Without a compelling operational or commercial rationale for backhauling Wi-Fi traffic to the Mobile Core, operators have instead leveraged local traffic break-out with secure SIM-based authentication, often paired with policy control from the Mobile Core. This approach avoids adding unnecessary load to the Mobile Core, allowing operators to enforce policies locally through advanced service management systems like the Enea Aptilo SMP.
Mobile device vendors also play a critical role in determining what capabilities are implemented and what functionality that gets deployed into real-life services (Learn more about this in our recent post Wi-Fi Offloading and the Device). For one, the device decides whether to connect to Wi-Fi or stay on Cellular, and it generally performs well in network selection. It took nearly a decade for device vendors to adopt the IPsec client required for untrusted Wi-Fi access. The motivation for this change materialized with the rise of Wi-Fi Calling, which served both the device manufacturers’ interests and the consumers.
Update Dec 2025: Our latest innovation, QoE-based Traffic Steering—finalist in the Wi-Fi Now Awards 2025—gives mobile operators precise control by transferring the final decision-making in network selection from the device to the operator.
Backhauling Wi-Fi Traffic to the Mobile Core: Control vs. Efficiency
In certain markets, some mobile operators choose to backhaul Wi-Fi traffic to the mobile core, mainly for perceived control and regulatory aspects. However, a vast majority (90%) prefer local traffic break-out due to cost and network optimization reasons. The preference for local traffic break-out is driven by the desire to minimize latency and reduce core network load, and it is possible to move control to the edge.
There are two primary reasons operators still consider backhauling traffic:
Deep Packet Inspection (DPI): By routing traffic through the mobile core, operators can perform deep packet inspection (DPI) to analyze and manage data flows. However, this can also be achieved through policy-based routing to a central DPI function. It’s important to note that this approach does not qualify as backhauling under 3GPP specifications.
Policy Control, Quota Management, and Charging: Although these functions are managed in the mobile core, there is no strong justification for backhauling the traffic. A well-integrated Wi-Fi service management system can handle policy enforcement, quota tracking, and charging locally within the Wi-Fi network while still maintaining alignment with the mobile core. When charging for the Wi-Fi service, operators can manage quotas at the network edge by checking-out time/data quotas from the BSS charging systems, with the ability to offer top-up options or return unused quotas. Explore our solutions for local policy and charging integration.
Learn More About Wi-Fi Offloading Architectures
In the next upcoming white paper excerpts, we will cover the 3GPP specified architectures for Wi-Fi Offloading integrated with the Evolved Packet Core (EPC) and the 5G standalone mobile core (5G SA). If you don’t want to wait for these posts, go here for further insights on all the different Wi-Fi offloading architectures, and to learn why a standard 3GPP AAA function is not enough for effective Wi-Fi offloading.
