One of the most prominent technologies in the first 5G deployments is the 3GPP-based EN-DC (E-UTRAN New Radio Dual Connectivity), which allows devices to connect to 4G and 5G radio networks simultaneously. This is a key element in 5G NR Non-Standalone networks, which still use an LTE core.
Last month, SK Telecom and Samsung completed an EN-DC device interoperability test which achieved 2.65Gbps data speeds to the 5G smartphone, by combining 1.5Gbps from the 3.5 GHz 5G band (100 MHz in bandwidth) with 1.15Gbps from LTE in 1.8 GHz, 2.1 GHz, and 2.6 GHz bands (65 MHz in total). The test used Samsung’s commercial 4G and 5G NR networks and its virtual core (vCore).
SK Telecom said it would be able to increase overall data speeds by 80% by leveraging dual connectivity. That will allow users within the 5G coverage area to download a UHD video of 2GB in size in just six seconds, it pointed out. There will be no break in the video if the user moves into a 4G-only area, though speeds will fall.
But challenges arise in deploying EN-DC efficiently when the 5G supplier is different to that of the 4G network. Of the three 3GPP-defined options for EN-DC (see inset), Option 3a does support multivendor networks, but some operators claim this is not efficient. For instance, Vodafone UK’s CTO, Scott Petty, recently said: “You can’t run one 5G vendor attached to another 4G vendor”.
Nokia recently proposed what it claims to be a more efficient approach to deploying a 5G RAN from one OEM on top of an LTE network from another. Clearly, Nokia will have had its eyes on operators which have 4G networks from Huawei or ZTE, but may now face restrictions on using those suppliers in their 5G deployments because of potential government bans over cybersecurity claims. The Vodafone-Hutchison joint venture in Australia is an example of a Huawei shop which has now been prevented from sticking with the supplier for 5G, something that it says will increase the cost and complexity of its transition considerably.
You can read the full blogpost here, which outlines Nokia’s solution. The main thrust of the proposition is that it does not require X2 interworking – X2 theoretically supports inter-vendor interface between a 5G NR radio layer and the LTE anchor layer, but it is one of those standards, like CPRI, which has actually been implemented differently by different vendors, making the interworking imperfect.
Nokia claims “a simpler alternative solution that does not require X2 interworking between an incumbent and new vendor but facilitates a ‘new’ Nokia overlay with minimal support from an incumbent vendor”. This involves an overlay – deploying a new LTE radio from Nokia in a lower band, along with the 5G NR in a higher band, and using the new 4G radio as the NSA anchor. When the operator moves to standalone 5G, the lower band spectrum can be refarmed for 5G coverage.
The legacy 4G network still carries most of the LTE traffic, but the anchor control functions are taken on by the new 4G radio. This would not require large amounts of spectrum, argues the blog author Harri Holma – typically 5-10 MHz. “Operators may have a new block of spectrum where this additional amount can be taken from, like the new 700 MHz allocation which is unused in many markets today,” he said. “Another option is at 2100 MHz, which is still partly used by 3G and could be refarmed to LTE/5G while moving 3G to 900 MHz.”
Nokia adds: “The additional capacity opened by adding new 5G spectrum, and the migration of users’ data traffic to that, means that the overall network capacity can be substantially increased even if a small amount of LTE capacity is used in this way by the Nokia overlay.”
And on the cost front, Holma told TMN: “An additional LTE RF unit is needed in the first phase but a low band RF unit is needed anyway for 5G with standalone architecture. Nokia’s multiband antennas mean that this additional LTE layer can be added in the same tower climb as would be required anyway when adding new 5G spectrum to the network. Therefore, we are saying that there are no additional investments needed for the LTE anchor because a similar RF investment is anyway needed for 5G SA.”
The three EN-DC options:
There are three ways to implement EN-DC in a 5G NR NSA network, of which Option 3x is proving the most popular. They are:
- Option 3. The 5G NR site does not have a direct (S1-U) user plane link to the 4G evolved packet core (EPC). The 4G site must therefore carry all the 5G NR traffic via the X2 interface, which may involve upgrading the 4G computing and backhaul capacity.
- Option 3a. This has a direct S1-U interface from the EPC to the 5G NR site, and the traffic flow for each 4G or 5G NR is split at the EPC. This reduces the processing demand on the 4G site and works well in multivendor scenarios.
- Option 3x. This allows 5G NR to be connected simultaneously to both the EPC and the LTE cell via the traffic plane S1-U and the X2 interfaces, respectively. This serves to reduce the need for backhaul and 4G hardware capacity upgrades. The control plane always terminates at the LTE cell. This is the optimal solution for MNOs where both 4G and 5G RAN are provided by the same vendor.