A new whitepaper from Ericsson says that E-Band spectrum (70/80Ghz) will be the key to meeting future cell site capacity demands.
The paper says that in 2020 65% of all cell sites – excluding the major Asian markets of China, Japan, South Korea and Taiwan – will still be connected with microwave. With the top 20% of sites demanding 1Gbps connectivity, Ericsson thinks that E-Band connectivity could account for 20% of all new deployments by 2020, with traditional bands still accounting for 70 percent.
It says that E-Band will often be bonded with a lower (traditional) band link, to provide a large capacity low-availability link alongside a high-availability link operating at lower throughputs but configured to ensure connectivity for prioritised traffic.
The whitepaper says:
“A radio link installation using traditional frequency bands is capable of up to 300–500 Mbps with high availability (99.999 percent). In urban areas the hop length is often less than 1–4 km. By adding an E-band radio link with 250–750 MHz channels and combining it with the existing radio link, while using QoS mechanisms for prioritization, it is possible to boost capacity by an additional 500 Mbps to 5 Gbps. Having done this, it is possible to achieve 99.8–99.99 percent availability while still securing 99.999 percent availability for high priority traffic. The annual average of available traffic capacity will reach very close to the maximum for the bonded links, typically around 96–98 percent of the theoretical maximum peak capacity.”
Fronthaul
Another application for E-Band discussed in the paper is fronthaul for C-RAN deployments. Here, Ericsson claims that lossless CPRI compression implemented by the RAN vendor (fronthaul must be controlled by the RAN not the backhaul link node) will mean that 2.5Gbps links will be enough to support a single sector site with carrier aggregation. Multiple-sector support may be more of an issue – as link bandwidths increase linearly – meaning a three sector site will require an under-utilised 10Gbps link.
However, the paper notes the likely future impact of 5G and NFV developing in tandem. 5G’s lower latencies will mean that the radio element will need to be close to the antenna, while NFV will mean some centralisation of functions. That could lead to an architectural change: “There will most likely be another type of internal interface between the central part and the part close to the antenna. This interface will not be a CPRI as it is known today.” This, combined with the ability to compress CPRI means the need for higher capacities in the frontgaul (ie above 2.5GHz) is not that urgent, the paper concludes.
