Does 5G Hold the Key to Unlocking the Edge?
September 24, 2021
Edge computing disrupts the classic client/server model by decentralizing and distributing computational tasks closer to the application data sources. This architecture requires compute & storage nodes at the network edge interconnected by a network that can support the application’s connectivity requirements. This may seem easy, but connection requirements can vary dramatically between applications.
For example, an analytics application may require a very high number of simultaneous sensor connections sending small amounts of data at regular intervals. Autonomous driving applications require a network with extremely low latency for quick analysis and reaction times. Ultra-high definition multimedia applications require a network that can support high bandwidth streaming services.
Enter 5G - the network designed to support high capacity, low latency, and high bandwidth applications.
5G Radio Access Network is Key
5G introduces three different spectrum bands that serve a specific purpose. This is key to being able to support the widely varying requirements of edge computing:
- Low-band (sub 1GHz). This is also the spectrum used by LTE today. It’s the best option to maximize coverage, but the maximum bandwidth tops out at about 100Mbps.
- Mid-band (sub 6GHz). Mid-band provides higher bandwidth to 1Gbps and lower latency which is critical to many IoT or machine-to-machine applications. However, reception through buildings and objects is a problem. 5G addresses this by using multiple input, multiple output (MIMO) macro cells to increase the number of simultaneous users. A technology called “Beamforming” is also a new innovation in 5G where the antenna sends a directed signal to every connected device with monitoring capability to increase signal quality.
- High-band (>6GHz) or mmWave. This gets the majority of the press when it comes to 5G because it provides peak data rates up to 10Gbps with extremely low latency. As you might expect, the price you pay is very limited distance (less than one square mile) and poor object/building penetration. It does however fit a nice sweet spot between technologies like WiFi and it’s low and mid-band predecessors.
5G RAN edge routing is another key feature. In prior generations, all user equipment traffic had to be routed through a centralized packet gateway. 5G enables traffic routing between devices and intelligent edge components or between devices which dramatically reduces latency and network congestion.
Edge Computing Hype-Cycle - “You are Here”
All this innovation is fantastic, so edge computing deployments and applications must be happening daily, right?
In order to get some real-world information, I spoke with Rajat Sethi (Vice President, Engineering Services) and Narendra Dhara (CTO) at ACL Digital. ACL Digital works with manufacturers, product companies, operators, and enterprises applying new and emerging technologies to help define roadmaps and next generation products & services. One of their key focus areas is 5G and Edge computing.
Narendra mentioned another key enabling factor is 5G virtualization and technology accessibility.
“Cost of deployment is going down due to open source consortiums. It’s possible to build 5G applications completely out of open source repositories. But then the problem shifts to maintaining, productizing, and hardening the solution.”
Narendra said that customers are prepared to ramp up their workforce and partner with experienced technologists in order to develop and manage these solutions as they scale. But these providers grapple with the CapEx/OpEx formula to optimize the cost model for their application.
Narendra and Rajat agreed that there are pockets of trials and deployments, but most of the focus right now is simply getting the ecosystem components where they need to be and the baseline edge computing infrastructure in place for 5G edge computing.
“There are some trials” Rajat mentioned. “But in many ways, we don’t even have a 5G network with all 3 spectrum bands, country-wide under one entity yet. All 5G RAN spectrums are not widely available making production roll-outs difficult or unrealistic. A lot of planning work is in greenfield opportunities which takes longer to define and develop. The brownfield implementations are planning their transition to 5G edge computing, but the new disaggregated software/hardware will have to coexist with legacy for a while hence, there are technology interoperability, system integration, testing, and verification at scale challenges that must be addressed”.
Narendra mentioned we should expect non-traditional players to enter 5G edge computing as well.
“Ease of adoption through open source has lowered the bar to other industry players. For example entities building and managing cell tower sites can now start thinking about putting 5G in a box and spinning up their solution for backhaul service. Independent enterprises and satellite infrastructure providers are also becoming increasingly involved”.
Rajat felt that there is a big opportunity for legacy customers to move their equipment toward 5G. Performing the virtualization and migrating equipment to 5G RAN is an important first step. This primarily involves network operators, but also requires key contributions from chipset, network equipment, and embedded systems companies as well.
What are you doing to contribute to the 5G edge computing ecosystem? Comment below or email [email protected]. I’d love to hear about it!