5G transport will require a new router at the cell site. There are two main drivers. First will be the increase in the amount of bandwidth needed as cell sites expand and more radios are added. With the change to eCPRI, fronthaul gateways will no longer be required. Second, the router must be able to support new services with more bandwidth and features such as MPLS, hierarchical QoS on ingress and egress, and segment routing for traffic engineering purposes.
We have noted before that there is no one size fits all for 5G transport. This is driven by changes to the cell sites that will not look or stay the same over the next few years. We have noted that 4G and 5G must coexist at the cell sites for the foreseeable future. We can use urban macro sites, urban small cell sites, and rural and suburban sites to describe how the changes to cell sites will drive changes to 5G transport and routing.
First, we see 5G RAN added to existing macro sites as operators begin their build-outs. Rather than adding a dedicated DU, we see the significant CapEx advantages of a virtual DU (vDU) running on COTS hardware at the cell site. We expect that the cell site router (CSR) function and the vDU can both run on this hardware. The CU would be located in a more centralized location (as would the 4G BBU if C RAN had already been implemented). The distributed vDU has the benefit of lower bandwidth requirements than the RU to DU link which saves money in the short term
Second, operators may opt to connect small cell sites through the macro site depending on their service area and fiber footprint. These can be consolidated on the same cell site router even if that router is running on an X86 server is shown here. Operators can scale the number of servers to match the number of cell sites to be connected.
Third, the macro site will expand with more 5G RUs. This densification which may be coupled with the aggregation of small cell sites is likely to exceed the capacity of an X86 server. As a result, we would expect operators to upgrade these sites to dedicated Ethernet switches. This densification will result in the centralization of the DUs which we would still expect to be deployed as vDUs.
Fourth, at suburban or rural cell sites, the density of 5G may not be as great as you would see in a dense urban area. In the US, one tower operator estimated that over 40% of their cell sites fell into this category. This represents a significant number of the cell sites, but they serve a relatively smaller population hence it was likely to be less dense. Rural cell sites are likely to have distance limitations that exceed the latency budgets between an RU and DU. Thus, keeping the vDU at the cell site is the least costly way to be able to implement 5G in these environments.
Finally, the Disaggregated Cell Site Gateway (DCSG) model is best suited to the aggregation of smaller cell sites. Small cell sites are likely to be deployed more broadly in dense urban environments and they lack the power, space, and cooling to be able to handle a significant amount of equipment. Using a DCSG to aggregate multiple small cell sites helps minimize the capital investment required at both the small cell site and the fiber plant. The distances involved are likely to be well within the budget of an RU to DU and as a result, the DU use can be collocated at a location like a central office in a centralized configuration.
5G transport networks will clearly evolve as 5G RAN expands and changes over time. A flexible and scalable routing platform will best meet these changing needs much more cost-effectively than traditional routers. Investments in service and automation can be preserved even as the underlying hardware changes. Operators can start small and pay as they grow.