Executive Summary

Patchy mid-band mobile network coverage will increase network transitions.

MSO’s must combine all of their assets to deliver services efficiently, without impacting user experience. Currently available solutions are lacking and will require the MSO to make tradeoffs between network efficiency and user experience.

The 3GPP is addressing some of the challenges in new releases, but implementation of those new specifications will take many years. Therefore, early engagement with the vendor ecosystem is needed to commercialise emerging technologies such as “Dynamic DSDS” and “Overlay ATSSS” which can help MSOs to achieve their goals now without dependance on MNOs:

Deliver seamless transitions when users are moving between areas of coverage,
Maximise device attachment to the lower cost network hPLMN and/or Wi-Fi

Patchy coverage of mid-band 5G networks

Patchy coverage of mid-band networks — Figure 1

With the availability of new spectrum resources, we see the advent of new mobile operators. In most cases, these operators (Multiple System Operators or MSOs) already operate fixed networks and provide mobile network services as MVNOs .

Typically, most relevant new spectrum for 5G is available in mid-band frequencies i.e. 3.5 GHz and even more spectrum is available in the 26 GHz range. This spectrum allows new operators to build their own networks but with only small cell radii creating patchy coverage. They will still rely on their MVNO / roaming contracts with hosting MNOs for coverage outside of busy areas covered with their own network.

In addition to those 3GPP networks, an MSO might also have Wi-Fi resources available to further optimise usage i.e. offload traffic from vPLMN (visited Public Land Mobile Network) and hPLMN (home Public Land Mobile Network).

When using both hPLMN and vPLMN networks to achieve maximum coverage, the MSO is motivated to:

Deliver seamless transition when users are moving between areas of coverage,
Maximise device attachment to the lower cost network hPLMN and/or Wi-Fi

On the other side, the MNO is motivated to maximise usage of its network by visiting customers and to minimise offload, and is therefore not incentivized to open hPLMN network and roaming interfaces required by an MSO to fulfil their goals mentioned above.

Moving between Wi-Fi networks, hPLMN and vPLMN creates lots of Edge-of-Coverage situations and subsequently network transitions which might negatively impact UX, so our question is:

How can MSOs optimise usage of their own resources to minimise network cost while maintaining user experience (UX)?

Mobility between networks and UX implications

Let’s observe what is happening with an MSO user as they make network transitions during their day (Figure 2). In our example, home routing (HR) mode will be used for roaming between hPLMN and vPLMN so users can access services via the Packet Gateway (PGW) of hPLMN and WiFi networks at home and public venues are added to complete the picture.

Implications of mobility between networks — Figure 2

What we want to focus on is what is going on in the Edge of Network situations when a user is moving in and out of areas covered by different networks. We stated earlier that the MSO wants to optimise usage of networks by preferring to keep the data traffic on Wi-Fi- > hPLMN -> vPLMN (order of preference) while maintaining high and consistent UX.

Let’s review what is happening with user connections and subsequently with UX in each of those Edge of Network situations as the user is moving through different areas of coverage.

1) User moves from home Wi-Fi into the vPLMN coverage area

Wi-Fi is the preferred connection on the UE (User Equipment), so the user will be stuck on Wi-Fi until the UE decides to release the connection. This is a bit different for different devices and dependent on current service usage. Still, since there is no connection between the Wi-Fi network and vPLMN, user connections to Internet services will be dropped and new connections will be re-established on the vPLMN. Of course, some services are more resilient than others to network interruptions. Even more problematic for UX is if the user is staying in Edge of Network areas where Wi-Fi attachment is still not released, but a poor Wi-Fi connection is severely impairing UX. In this situation, the only remedy is that the user manually switches off Wi-Fi in order to force attachment to the vPLMN. If this happens, the user will often forget to switch Wi-Fi back on later and will stay on the vPLMN even in situations where good Wi-Fi service is available.

2) User is on vPLMN and is approaching a city centre with good vPLMN coverage and enters an area where hPLMN coverage is available

When the user is approaching an hPLMN coverage area, we want their connection to transition from vPLMN to hPLMN as quickly as possible and without interruption. How and when this happens, will depend on the roaming contract between MSO and MNO and the implementation of various parameters in both networks. ¹ The key for seamless handover is to include the S10 interface in the roaming agreement. S10 will establish coordination between a vMME and hMME (Mobility Management Entities) and enable seamless transition in both Idle (RRC_IDLE) and Connected (RRC_CONNECTED) modes. Once this interface is available both MMEs can negotiate seamless cell reselection or handover between neighbouring cells of vPLMN and hPLMN, also favoring attachment to hPLMN which will be set as a higher priority network.

Since MSO and MNO have conflicting interests in roaming agreements, S10 is typically not part of the agreement, so seamless cell reselection and handover are not possible. As a result, network attachment to the hPLMN will not be optimised and network transition will not be seamless. In Idle mode, the user will stay attached to the vPLMN until a higher priority network is detected (up to 6 minutes). For Connected mode, the user will stay connected to the vPLMN until the radio link failure is detected, after which the UE needs to search for the higher priority network (hPLMN) and complete network authorisation (800-900 ms). In both cases it is obvious that the MSO goal of maximising attachment to its own hPLMN can not be achieved.

Results and UX impact from typical configuration tested by Cable Labs ² is given below. In this table, similar results to those measured for the LBO (Local Break Out) case are also obtained in the case of HR (Home Routed) where the S10 interface is not implemented.

Table comparing the network implementations of Home Routed with Local Break Out across various metrics — Figure 3

3) User is attached to the hPLMN and is leaving the hPLMN coverage area

When the user is leaving an hPLMN coverage area similar considerations will apply to handovers as in the previous case (assuming no S10 interface between the networks), but triggers for network change will now be under the control of the MSO. The MSO can configure signal level for the start of the transition from the hPLMN to the vPLMN and inform the UE about the preferred network and frequency to attach to. This will shorten the time required for the UE to reattach to the vPLMN (redirection). This was not the case when the user was entering hPLMN coverage, because the vPLMN would need to set those parameters. Results in the table above are also depicting this case.

4) User is attached to the vPLMN and moves into a public venue with Wi-Fi coverage

When entering a venue with a known Wi-Fi network, the UE will attach to Wi-Fi and use Wi-Fi as the preferred Internet connection. There will be no “handover”, but transition will be made gracefully. In case of no active data session, transition will happen as soon as Wi-Fi is in range and in case of an active data connection on vPLMN, the active session will finish on vPLMN and a new session will start on Wi-Fi. When leaving public Wi-Fi, the use case described under 1) applies.

As we can see from different use cases, the MSO does not have all the levers to efficiently steer the customer to the preferred network while keeping the UX high and consistent.

Key challenges are:

Stickiness to poor Wi-Fi at the Edge of the Wi-Fi network and handover from Wi-Fi to vPLMN -> UX impact

Stickiness to vPLMN when entering area of dual coverage (vPLMN and hPLMN) -> Network cost impact

Handover from vPLMN<->hPLMN in cases where the S10 interface is not implemented in the roaming agreement -> UX impact

Possible solutions

Solutions for identified problems can be found within future 3GPP releases, but also outside of 3GPP specified solutions.

3GPP is addressing multipath connectivity between non-3GPP and 3GPP networks by implementing the ATSSS (Access Traffic Steering Switching and Splitting) function as a part of the UPF. The first ATSSS specification came out with 3GPP Rel 16 in 2020, but it was not picked up into implementations by UPF vendors, so to date no ATSSS implementation in UPF is known to us. The upcoming 3GPP Rel18 is expected to include additional major enhancements in ATSSS functionality, but it will take years for Rel 18 based UPF’s to be implemented and then to find their way into operator networks.

Considering that the underlying multipath technology is available and commercially deployed in other use cases (such as Hybrid Internet Access), Tessares, as an early contributor to ATSSS technology³, proposed an “above-the-core” ATSSS function in order to decouple the timetable for ATSSS from slow moving implementations of 5G SA Core and UPF vendors.

In 2021, the first operator trials showed promising results and we are encouraging more operators and UE manufacturers to get on board in 2022. Feedback from the trials will shape the technology in the drive towards commercialisation.

Tessares Overlay ATSSS is network agnostic and helps improve UX during Edge of Network situations by creating so-called “make-before-break” handovers independent of 3GPP network signalling. This means that connections will be temporarily split over 2 paths before one path is lost and in that way the user connection is not lost during handover between networks.

Comparing downloads where Wi-Fi and LTE are available but with or without ATSSS — Figure 4

A similar effect can be achieved in the Edge-of-Network situation between hPLMN and vPLMN, but the difference in this case is that the UE can establish only one connection towards the Internet (either through vPLMN or hPLMN) and ATSSS can not impact network attachment triggers.

Comparison of the network efficiency advantage of using ATSSS to achieve a seamless network handover. — Figure 5

In order to tackle the problem of “vPLMN stickiness”, the MSO can use UEs with enhanced DSDS (Dual SIM Dual Standby) functionality as proposed by Charter in this paper⁴. Technology like Dynamic DSDS or Opportunistic DSDS will add additional logic to standard DSDS UEs. This will enable them to dynamically change primary network between vPLMN and hPLMN based on measured network parameters and location data. The UE will not use the vPLMN more than necessary. As a consequence of this, MSO customers will experience much more network transitions than MNO customers, so those network transitions have to be seamless for users. This is where Tessares overlay ATSSS comes in.

5G Core implications

In the use case of the MSO with mid-band 5G PLMN, users will be mainly attached to the 4G vPLMN network macrocell while in the vPLMN coverage area because the area that the MNO will cover with 5G NR small cells is roughly the same area that the MSO will cover with their own network. As long as both interfacing networks are not 5G SA, handover procedures will be similar to LTE roaming and limitations to UX and network steering to hPLMN from previous considerations will remain unchanged.

In general, 5G does not bring many changes to mobility aspects and this is particularly relevant for roaming. All considerations discussed for 4G apply to 5G also. Some protocols and interfaces will be different (S10->N14 ; S8-> N9&N8 etc), but the concept and resulting impact on UX are comparable.

5G network architecture for home routing while roaming — Figure 6

Conclusion

Considering patchy coverage of mid-band based mobile networks and resulting increased number of Edge of Network situations, MSO’s have to think about how to combine all of their assets to deliver network services efficiently, but without impacting user experience. We see that currently available solutions are not ideal and will require the MSO to make tradeoffs between network efficiency and user experience.

3GPP is addressing some of the challenges in new releases, but implementation of those new specifications will take many years. Therefore, early engagement with the vendor ecosystem is needed to commercialise emerging technologies such as “Dynamic DSDS” and “Overlay ATSSS” which can help MSOs to achieve their goals now and without dependance on MNOs: