The quality, and resulting user experience, of mobile data applications depends on two key performance indicators: high-speed data throughput, and low latency. 5G network designers are taking advantage of large spectrum swaths in mid-band and millimeter wave frequencies to increase data throughput speeds close to 1 Gbps. Network densification with small cells, fiber and edge computing together help reduce latency to low single-digit milliseconds (ms).
Not all applications require the same high-speed data and low latency performance, however. Certainly, every wireless application involves a mix of radio access network and transport network elements over some distance to reach servers in a data center.
Delivering high data throughput is a function of RAN operating spectrum and available channel bandwidth, antenna proximity to user equipment, massive MIMO antennas with beamforming and fiber optic cable distance to core switching and routing functions.
The need for speed is escalating. Each generation of wireless technology provides roughly an order of magnitude improvement in data throughput that enables greater public and private internet usage.
Video streaming and multiplayer gaming are the latest killer apps. The Ericsson Mobility Report (Nov 2020) projects that global mobile data traffic per smartphone will increase from an average of about 10 Gb per month in 2020 to around 35 Gb per month by 2026 with some regions such as North America, Western Europe and North East Asia reaching 40-50 Gb per month.
Why is latency important?
Latency is the lag or delay in reaction time to a stimulus or transaction. Research indicates that humans have an average reaction time of 200-250 ms or about a quarter of a second (1 second = 1000 milliseconds). (Think you are quick? Test yourself here!).
End-to-end (E2E) transport latency in the ms range varies with how far signals must travel in a network from user equipment to a data processing point, how many routers they pass through, whether the connection is wired or wireless, and the type of application.
For instance, delays occur when connecting to web sites from wired and wireless devices. Most people will tolerate a second or two delay to access a site. Enhanced mobile broadband and massive machine-type communications are not so forgiving.
The typical latency of a 4G LTE network is about 50 ms. VoIP and video calling such as Facetime or WhatsApp have latencies of tens of ms.
5G is promising low latencies for applications where even small network delays can be problematic. Fast, multiplayer interactive games such as drone racing require 20–30 ms end-to-end network latency. Autonomous vehicles and factory automation require latencies of 10 ms or less.
Virtual reality, augmented reality and haptic communications or communicating a sense of touch all require latencies in the single digit ms range for applications such as teleoperating robots or even telemedicine involving remote surgical procedures.
Network designers are addressing data throughput and latency factors to meet a growing range of applications that 5G enables.
5G RAN elements connect via fiber to the network core that is virtualized in data centers. How far data centers are situated from the RAN determines the network configuration needed to support user applications at a particular location.
Fiber transmission is at the speed of light which, if you do the math assuming 300 million meters per second, is 300 km/ms through the air and is estimated at roughly 200 km/ms in glass fiber.
Latency in connecting to global data centers used by web hosting hyperscalers like AWS, Azure and Google is roughly 100 ms plus inherent delays in routers and UEs. Connecting to regional data centers that host both public and private cloud services lowers the latency to the 20-30 ms range.
Specialized applications requiring 5G ultra-reliable low-latency communications in the single digit or even sub-1 ms range will require some sort of edge computing at or close to cell sites.
Depending on the performance specified for a given application, the associated capital expenditures and operating expense will factor into which network design is selected.
By John Celentano, Inside Towers Business Editor