Advanced Computing in the Age of AI | Thursday, March 28, 2024

Intel Rolls Out its IoT Platform 

The Internet of Things is perhaps transitioning from catchy buzz phrase to an organizing principle and, to hear one emerging player tell it, an evolutionary step in computing.

Chip giant Intel Corp. pulled out most of the stops this week to showcase its year-old IoT initiative embodied in what it calls the Intel IoT platform. The emerging ecosystem exhibits characteristics the company deems critical: the ability to scale and, according to Doug Davis, general manager of Intel's IoT initiative, a "repeatable foundation" that will enable the development of connect products and accompanying data services.

Intel_IoT_Platform

The explosion of connected "things" (Intel estimates 40 billion sensors and 10 billion devices by 2020), an estimated 40-fold increase in the wireless device connections along with the Intel's founding principle—Moore's Law—are all driving the Internet of Things. That translates into inestimable amounts of new sensor data that must be aggregated and stored. Hence, datacenters and cloud management are integral to supporting and scaling the IoT as it moves from a marketing concept to a business with actual revenues.

Speaking of which, Intel claims about $2 billion in revenues over the past in year from its nascent IoT business. It trotted out several customers during a company event on Dec. 9 in an attempt to put flesh on the bones of its IoT infrastructure. These examples of the "value added" IoT capabilities included a New York City building management company with 10 million square feet of office and hotel space in Manhattan. It is using sensor data to regulate heating and cooling, a major cost center for commercial landlords, especially in the Big Apple.

John Gilbert of Rudin Management said utility data has long been available to property managers, but Rudin's IoT investment was the "catcher's mitt" that allowed it to leverage power data to transform grand Manhattan properties into smart buildings that use less power, about $1 million in annual energy savings by Gilbert's reckoning.

Intel used the building management example to illustrate how its IoT platform represents a "gateway" to access the coming fire hose of sensor data. Among the other IoT building blocks are device connectivity and provisioning provided by a range of increasingly "invisible" Intel processors on the front end and data "normalization" and cloud management on the backend, the company envisions.

Underlying its vision is a heightened state of network security in what Intel Labs managing director Wen-Hann Wang called the "Era of Ubiquitous Intelligence." In order to prevent others from knowing what you had for breakfast, Wang said Intel has been working on a "lightweight encryption" scheme called Enhanced Privacy Identification technology. Intel's version is now available for licensing. "Lightweight" refers to the relatively small amount of circuitry required to deploy the encryption scheme.

Meanwhile, the "connective tissue" that will link IoT devices with datacenters are APIs that not only provide secure provisioning of those connections but also would allow Intel's IoT platform to scale, according to Doug Fisher, Intel's general manager for software and services.

In order to ensure the network compatibility that will be needed for connected devices, Intel has helped form an Open Interconnect Consortium while jumping recently onto the OpenStack bandwagon. Fisher stressed that API management is among the ways Intel can scale its IoT platform.

Intel_IoT

Meanwhile, the chip maker is betting that wearable devices will be among the "things" that must be connected to its IoT platform. That, said Diane Bryant, general manager of Intel's Data Center Group, and big data analytics are "the next stage of computing." Datacenter hardware in the form of storage and computing will account for much of the estimated $40 billion in future cloud and edge networking revenue, Bryant predicted.

To cash in on the opportunity, Intel's IoT platform will require a single architecture for data analytics ranging from predictive to deep learning, Bryant added.

Among the new offerings emerging from Intel's cloud management structure is a projected data-as-a-service scheme, the company said.

EnterpriseAI