3G to LTE Handover – Another Approach

Back in 2014 I was musing about that the time has come for network operators to introduce a feature to hand-over an ongoing 3G connection to LTE. For me and those who frequently tether their PC while on the move, this is an important feature when the network makes a handover of an ongoing connection from LTE to 3G. Without the feature a device is stuck in 3G unless the radio bearer is released and the mobile can reselect back to LTE autonomously. Another way back to LTE is when the connection is transferred by the network to 2G at some point as here, the mobile can take measurements on LTE bands autonomously even while a data transfer is ongoing and reselect back to LTE. Unfortunately not much has happened in in that regard since then and I continue to be stuck on 3G for a long time when using my notebook. But when I was recently traveling I observed a number of 3G to LTE ‘handovers’ while a data transfer was ongoing for the first time. And it worked a bit differently from what I expected.I so far assumed that once I would see the feature in the network it would be performed with an inter-RAT handover. To do that the network would have to introduce communication gaps by activating 3G compressed mode for the device once it thinks there is a chance that the device can receive an LTE signal and then configure E-UTRAN measurements. This could be done periodically if 3G and LTE coverage is mostly overlapping or based on measurement reports, i.e. if the signal level gets too low at the cell edge or the signal level goes beyond a certain threshold, i.e. the user moves closer to the center of a 3G/LTE cell. Once the mobile reports back that an LTE signal can be received with sufficient quality the network would then perform an inter-RAT handover. So far the theory.

In practice, however, I observed a different procedure. Instead of a full handover, the network sent a ‘Directed Signalling Connection Re-Establishment to EUTRAN’ command. In other words, the 3G connection was not handed-over but instead terminated and re-established in LTE. In practice this is obviously not as good as a handover. However, since connection establishment in LTE only takes around a hundred milliseconds it still works quite well. For the network the procedure is a lot simpler as the 3G and LTE sides do not have interwork tightly from a signaling connection point of view. For example, this way it is not necessary to reserve resources in the LTE eNodeB prior to the transfer which makes life a lot simpler.

In practice I was quite happy to see this happening as my user experience, especially during tethering when IP packets more or less flow constantly due to a lot of web based applications, was significantly improved. Unfortunately it is still rare to observe this feature in practice. But perhaps now that it’s in one network perhaps other network operators who strive for good network quality will finally follow.

P.S.: The network will only configure measurements if the mobile needs is able to measure LTE cells while in UTRA compressed mode and tells the network of its capabilities. In 2017 I assume that most mobile devices do have this capability. It does this by setting bit 2 of the optional EUTRA Feature Group Indicators during the RRC connection establishment procedure (EUTRAN measurements and reporting in connected mode). Have a look at 3GPP TS 25.331, Annex E.1 for the details.

2 thoughts on “3G to LTE Handover – Another Approach”

  1. In case of RRC CONNECTION RELEASE message with cause “Directed Signalling connection re-establishment”, will the user’s IP address (assigned in 3G GGSN) remain the same? It probably depends on operator’s policy, but cannot be guaranteed.
    Most operators assign IP addresses dynamically, some operators offer subscription with static IP address (at extra charge). Operators may assign IP addresses in semi-static fashion: the subscriber gets the same address as previously, unless long time passed since last PDP context activation or shortage of addresses occurred while the subscriber was detached.

    When user’s IP address changes, most web services will discontinue https connection (for security reasons). In case of ‘real’ handover the IP address is expected to remain the same, because the gateway to internet is anchored.

    Would unchanged IP address matter to the subscriber?

    1. The procedure is similar to just changing between 4G and 3G and you keep the IP address in that procedure as well. Never saw a network that threw the context away during that.

Comments are closed.