On the back of last night’s unveiling of its enterprise Cloud Services Platform, Google has just announced two new products aimed at helping customers develop and deploy intelligent connected devices at scale.
Edge TPU is a new IoT-focused chip for the edge environment, and Cloud IoT Edge is a software stack that extends Google Cloud’s AI capability to gateways and connected devices.
Together, the two products allow users to build and train machine learning models in the cloud, then run them on Cloud IoT Edge devices via the Edge TPU hardware accelerator.
Adding greater intelligence and speed at the edge will be critical in the development of IoT networks – such as smart city and connected transport initiatives, for example.
Into the detail
Google has provided further details on the products. Edge TPU is a purpose-built ASIC chip designed to run the company’s TensorFlow Lite machine learning (ML) models at the edge – in short, a ‘lite’ version of its Tensor Processing Unit coprocessor family.
“When designing Edge TPU, we were hyper-focused on optimising for ‘performance per watt’ and ‘performance per dollar’ within a small footprint,” said Injong Rhee, VP IoT at Google Cloud.
“Edge TPUs are designed to complement our Cloud TPU offering, so you can accelerate ML training in the cloud, then have lightning-fast ML inference at the edge. Your sensors become more than data collectors – they make local, real-time, intelligent decisions.”
A possible application is analysing video feeds without having to shuttle data backwards and forwards to/from Google’s cloud, or an on-premise cloud solution (now available from Google, according to its cloud platform announcement yesterday).
The TPU will be made available to developers by October, said Google.
Meanwhile, Cloud IoT Edge extends Google Cloud’s data processing and machine learning capabilities to gateways, cameras, and end devices, making IoT applications smarter, more secure, and more reliable, according to Google.
“It lets you execute ML models trained in Google Cloud on the Edge TPU or on GPU- and CPU-based accelerators,” explained Rhee. “Cloud IoT Edge can run on Android Things or Linux OS-based devices,”
Its key components include:
- A runtime for gateway-class devices with at least one CPU, to store, translate, process, and derive intelligence locally from data at the edge, while interoperating with the rest of the Cloud IoT platform.
- The Edge IoT Core runtime securely connects edge devices to the cloud, enabling software and firmware updates and managing the exchange of data with the Cloud IoT Core.
- Meanwhile, the TensorFlow Lite-based Edge ML runtime performs local machine-learning-based inference using pre-trained models, reducing latency and increasing the versatility of edge devices.
Businesses can benefit by bringing machine learning to the edge in several different ways, said Rhee, including increased operational reliability, faster real-time predictions, and increased security.
Google is working with a range of partners on these new initiatives, including semiconductor providers NXP and ARM, gateway device makers Accton, Harting, Hitachi, Vantara, Nexcom, and Nokia; and edge computing specialists ADLINK Technology, Kelvin, Olea Edge Analytics, Smart Catch, and Trax.
Internet of Business says
Google’s move into edge hardware and services sees it coming late to a party that has only just started to swing. Microsoft, Dell, and others have been pouring billions of dollars into the IoT and edge environments – where speed, low power usage, and embedded intelligence will be critical factors in IoT services and networks.
For speed-critical IoT applications, processing data at the edge avoids the need to send data to and from remote data centres.
This is yet more evidence that the world’s number three enterprise cloud provider is serious about its ambitions to both move higher up the value chain, and closer to the point of need.
As the cloud becomes the enterprise’s computer, trust will be the key, and in this sense Google and Amazon may face a tougher battle than, say, IBM or even Microsoft, particularly as Google has opposed California’s recent data privacy rulings.
Read more on the the shift from cloud to edge computing.