The nation’s largest carriers are demonstrating the power of fifth-generation telecommunications technology to reshape mobile computing as it’s rolled out across the US. By processing information at faster speeds than remote data centers, the potential of 5G to disrupt cloud computing business models is on show, too.
In testing last week, engineers from Verizon routed 5G traffic through a local server to halve the time needed to execute a facial-recognition application compared with using a public cloud to process the data. AT&T, which runs a close second to Verizon in subscriber numbers, is also debuting a technology called multi-access edge compute (MEC) as it begins 5G service in a dozen cities across the US.
AT&T, Verizon and other carriers are leveraging MEC infrastructures and software platforms to process and store data closer to the network edge from which applications run. MEC relies on software-defined networks (SDNs) for the network virtualization that guides traffic between local data centers and the mobile devices and low-power sensors that comprise the Internet of Things.
The time saved in comparison to round-tripping traffic over multiple network relays to centralized server farms is critical for low-latency applications. These include virtual reality and self-driving vehicles, both of which are enabled by 5G’s wider frequency spectrum. In September, Deutsche Telekom, parent company of the number-three carrier T-Mobile that services 80 million US customers, partnered with Silicon Valley software design firm Aricent to develop an as-a-service MEC platform for 5G.
Microservers Cut Latency Times
The technology also opens the door to the localized provision of analytics, streaming and machine-to-machine communication services currently dominated by cloud computing giants including Amazon Web Services, Microsoft Azure, Google Cloud and IBM. As carriers build-out their 5G networks with the thousands of small-cell antennas necessary for handling low-power transmissions, they will look to install microservers in the hubs that connect them.
Verizon’s testOpens a new window took place in Houston, one of four US cities where the carrier began piloting 5G last October. They used the expanded 5G transmission spectrum to send data for processing both locally, in a radio access network hub as well as in a remote data center.
Verizon officials said that moving the back-end compute functionality closer to the application interface took round-trip speeds down to around 15 milliseconds. The increased speed permitted facial recognition to take place at more than two times the rate of a centralized server, with the lower latency cited as critical to the effectiveness of the application and others like it.
In late 2017, Verizon onboarded caching tech from a Silicon Valley MEC vendor called Qwilt that lets the carrier store and stream video at the network edge. By delivering content from physical locations closer to its customers, Verizon said network traffic flows were cut by 20%.
Prioritizing Application Traffic Flow
AT&T, which serves 153 million subscribers, is piloting MEC at Chicago’s Rush University Medical CenterOpens a new window in a trial that began in January 2019. The carrier’s SDN manages traffic through on-premises hardware, boosting processing speeds for medical imaging, patient sensors and robotic operations.
More efficient data flow management means that Rush administrators can direct applications-specific processes that call for low latency and high bandwidth to its local network infrastructure. Meanwhile, operations in which latency is less a factor such as records processing and storage can be routed to a public cloud.
T-Mobile’s parent company cites research that predicts the market for edge computing to rise to $3 billion by 2025. As Deutsche Telekom awaits approval from US regulators for its merger with Sprint – the No. 4 US carrier with 53 million subscribers – the German giant is conducting trials for its SDN at its Hubraum incubator in Berlin.
Unlike Verizon and AT&T, Deutsche Telekom’s partnership with French-owned Aricent is building an open-source code for network virtualizationOpens a new window . The goal, the company says, is to give developers tools to build-out use cases that boost the uptake of low-latency applications.
Currently serving 154 million US customers, Verizon’s optimism is high, even as use cases for what the company is calling its “Intelligent Edge Network†remain relatively few in number. The hope is that edge computing will drive innovation, with the growth in low-latency applications underpinning demand for 5G-enabled MEC.
With the pace of 5G rollout set to increase in the coming years in the US and around the globe, analysts say the market for vendors of devices from smartphones and sensors to the network hardware needed to carry voice, data and machine communication appears robust – as are the opportunities for telcosOpens a new window to deliver cloud-type services at the network edge as they recoup the investment in their 5G infrastructures.