Monday 26th August 2019

The leading edge: How IoT will drive a distributed computing architecture

Published on June 4th, 2019

Cloud computing has transformed the IT landscape, bringing massive benefits in costs, flexibility and reliability.

It is now the default approach for most applications, with the Cisco Global Cloud Index predicting that cloud processing will reach 94% by 2021 and Gartner forecasting that cloud will account for 28% of enterprise IT investment by 2023. This will represent some US$1.3 trillion (€1.1 trillion) of spend, Justin Shields of Vodafone and Zak Doffman of Digital Barriers say.

Meanwhile, cloud-based artificial intelligence (AI) has seen a parallel revolution, with open platforms providing easy access to powerful new capabilities such as image analysis, data insights and predictive maintenance.

The IoT data deluge

Yet just as cloud computing and cloud AI have centralised computing, so the Internet of Things (IoT) is set to redistribute it.

IoT today is a race to connect billions of physical devices into new applications that will transform industries and, indeed, our lives. IHS Markit recently forecast 125 billion such devices by 2030, up from 27 billion last year in areas such as security, healthcare, transport and manufacturing.

Justin Shields

As connected devices proliferate, cellular data growth will double from 25% to 50% per annum. Security video traffic in particular will see a seven-fold increase over the next four years, fuelled by the shift to cloud video platforms. New video analytics capabilities such as facial recognition, object classification and behavioural analytics provide further impetus to this growth.

This firehose of data will stretch even 5G cellular capacities. Our first is to keep the benefit of connected things while lowering the load they put on the network.

Driverless cars driving the change

Autonomous vehicles create a different imperative. Bristling with sensors, they measure everything they can to create a three-dimensional model of their surroundings and drive their occupants safely and efficiently through it.

Video, LIDAR, radar and ultrasonic sensors provide information about the environment. GPS, accelerometers and vehicle sensors provide information about the car itself. Serious computing power is required to interpret these inputs, particularly to perform real-time video analytics on multiple camera feeds. Then, another layer of AI is required to complete the control loop, with the primary self-driving outputs of acceleration, braking and steering.

This safety-critical control loop has almost no tolerance for delays; cellularly-connected cloud services cannot be part of it. Our second challenge is therefore to provide local area autonomy within a wide-area-connected subsystem.

Edging your bets

Zak Doffman

These two challenges drive the move to distributed architectures and an AI-enabled edge. Local intelligence delivers local autonomy and, by turning video into metadata, transformational reductions in bandwidth. Meanwhile, and right on cue, new neural processing hardware and algorithms are making device-side deep learning a practical proposition.

We still need the cloud. Distributed intelligence and data needs central coordination and no edge device can compete with the cloud’s capacity for data storage and processing. We need a fusion of cloud and intelligent-edge computing rather than reliance on one over the other.

We are at the early stages of evolution for this IoT architecture and it’s exciting to consider how it will evolve, as well as its impact on the wider computing landscape.

The authors are Justin Shields, Vodafone Business Products and Solutions director and Zak Doffman, founder & CEO of Digital Barriers.

About Justin

Justin plays an instrumental role to our success by defining and implementing the Technology strategy for Vodafone Business, managing all aspects of platform and product development together with Vodafone Business customer solutions, from design to delivery. Working closely with Vodafone Local Markets and our Vodafone Business Units, Justin and his team are focused on creating differentiated customer experience across the key priority areas underpinned by world-class service and the IT digital transformation.

Justin joined Vodafone in 2014 as the Group Enterprise IT Director for Enterprise, additionally picking up Enterprise IT in Vodafone UK in 2015. During this time, Justin has significantly improved the Global IT architecture across all elements of the Enterprise Sell, Build, Run model driving greater customer centricity, digitisation, innovation and simplicity.

Prior to joining Vodafone Justin led several major IT and Business Transformation programs across the Telco sector.

Comment on this article below or via Twitter: @IoTNow_OR @jcIoTnow