5G access requirements: Comparing the requirement for 5G access with fibre to the home and business access, there are some important differences. First of all, there will be strict delay requirements for 5G. Applications like autonomous cars and augmented reality dictates latency in the 1 millisecond range, but mobile fronthaul puts even stricter requirements to delay: Maximum 100 microseconds one-way delay between the Remote Radio Head (RRH) and the Base Band Unit (BBU). Secondly, the required capacity is likely to be 10 Gb/s or higher like 25 Gb/s, matching e.g. the bitrates defined in the eCPRI mobile fronthaul specification. Hence, both the delay and capacity requirements are fundamentally different from the access network we have today. Because of the high capacity required, also the cost per access point must be expected to be higher. These parameters dictate the suitability among available access techniques.
High availability of fibre resources: Comparing the different access techniques, it all comes down to cost and availability of fibre resources. If there are many fibres available at a low cost, dedicated fibres to each access point is typically the most cost effective solution. This solution is without compromises since it offers both the highest capacity and the lowest delay.
One important reason for point-to-point fibre becoming highly available is that fibrecables has become very cheap and compact. E.g. comparing with the diameter of a Twisted Pair (TP) cat-6 cable (approx. 6 m.m.), the fibrecable may contain e.g. 48 fibres and costs less than 4 Euro/meter. Hence, there is no reason saving on the fibre-pairs when digging down fibre for a few kilometre of distance.
Figure illustrates high availability of fibre resources allowing point-to-point dedicated fibre connections between BBU and RRH.
Scarce fibre resources: However, if fibre resources are scarce, the cost situation becomes different and it becomes attractive sharing the capacity in the fibre among several access points. This may happen e.g. if the network was built in the early days of fibre when putting down few fibres was not unusual, or if a mobile operator is renting fibres because they are not the owner of the physical access infrastructure. A third example is fibre-cables spun on top of electrical power-cables. Typically the maximum number of fibres in these cables is strongly limited.
The various Passive Optical Network (PON) techniques enable use of one fibre or fibrepair to a splitting point that may be placed close to the subscriber, splitting the capacity for sharing among several access points. Hence number of fibres is reduced in parts of the network. The PON may either split the capacity of a fibre by offering wavelengths to each access point (WDM-PON), or by using time division multiplexing (e.g. like GPON variants) offering time-slots to each access point, or both techniques may be combined. An additional benefit with GPON is that it enables savings in the number of transceivers at the line side, since one transceiver is shared among different access points. For meeting the ultra-low latency and the 10 Gb/s (or higher) capacity requirements for 5G, WDM-PON is a good candidate. A dedicated wavelength to each access point gives the lowest delay and meets current and future capacity requirements. This is in difference from GPON where capacity is shared, giving lower capacity and higher latency to each access point.
Figure illustrates WDM-PON applied for sharing the capacity of a fibre. Dedicated wavelengths may then be applied for connecting the BBUs and RRHs.
Sharing resources: Switching, aggregation and combining services: For increasing utilization further, sharing resources through layer 2 or layer 3 aggregation and switching methods may be cost effective. This allows aggregation of 10 Gb/s fronthaul access into e.g. 100 Gb/s wavelengths, and the combination of fronthaul and backhaul services within the same link or wavelength. Because of the ultra-low latency requirements of 5G fronthaul networks, special Time Sensitive Networking (TSN) switching techniques may be required in the fronthaul. Backhaul services on the other hand puts lower requirements to delay. For efficient resource sharing the backhaul may be statistical multiplexed with a QoS prioritized fronthaul service. This allows both high resource utilization and the fronthaul services to experience a lower delay than backhaul services, meeting the strict requirements.
Figure illustrates aggregation and deaggregation using low-latency Time Sensitive Network (TSN) switches. Ethernet switches may be applied for aggregating/deaggregating point-to-point fibre connections and/or WDM-PON wavelengths and/or microwave connections between the BBUs and RRHs .
If fiber resources are not available: Fibre resources may not be available at all, or being very expensive to deploy. This may be e.g. in a city between buildings where digging up the street is very expensive. It may also be to access remote areas with a hotspot. Wireless microwave links or optical wireless links may then be the solution. Using microwave links, 2.5 and 10 Gb/s capacities are available over kilometre distances. For optical wireless, directional access allowing high density between access points and data-rates beyond 10 Gb/s are easily achievable. However distance should be kept low (e.g. less than 100 m) because of the limited sight in fog, snow etc. The more limited the capacity, the more important it becomes to share the capacity efficiently, motivating the use of Ethernet switching or IP/MPLS routing solutions.