US network operator Verizon has completed its first test of edge computing on a live 5G cellular network, claiming that shifting computation closer to the user equipment has led to a halving of the network latency.
A key feature of 5G cellular networks, on top of improved bandwidth, is the promise of lower latency. With reduced latency compared to previous-generation cellular technologies new functionality becomes possible, from live drone piloting at range to virtual and augmented reality untethered from a bulky host machine.
5G alone, though, is unlikely to be enough to bring latency down to the point where this functionality becomes truly seamless. It’s here that researchers, including those at Lime Microsystems, are working on a potential solution: edge computing, bringing computation away from remote data centres and closer to the user equipment which relies upon it in a bid to reduce network load and improve latency – just as Verizon has demonstrated at its Houston 5G testbed facility.
“For applications requiring low latency, sending huge quantities of data to and from the centralised cloud is no longer practical. Data processing and management will need to take place much closer to the user. MEC [Multi-access Edge Compute] moves application processing, storage, and management to the Radio Access Network’s edge to deliver the desired low latency experiences, thereby enabling new disruptive technologies,” explains Adam Koeppe, Verizon’s senior vice president for network planning, of the trial. “This shift in where the application processing occurs, the inherent capabilities of 5G to move data more efficiently, and our use of millimetre-wave spectrum is a game-changer when it comes to the edge computing capabilities we can provide.”
Verizon’s test saw the time-to-recognise of a 5G-connected facial recognition system cut in half by using edge computing capabilities in place of the traditional remote data centre system – and shows potential for further improvement. “To achieve near-zero latency, where data moves many times faster than the blink of an eye, having computing functions closer to the user is a vital step,” adds Koeppe. “With this test, we have shown how much of an impact the move towards a MEC-based network architecture can make.”
Verizon’s MEC is one of a range of edge-computing systems currently being trialled or deployed, including the ultra-low-cost open-source LimeNET Micro, which brings the technology down to a price point suitable for educational and hobbyist use, and the CrowdCell project on which Lime Microsystems has partnered with Vodafone to develop and deploy lost-cost cellular-capable devices combining software defined radio (SDR) with general-purpose processing (GPP).
More information on Verizon’s edge computing trial is available on the official website.