JEDEC Forum: AI/Deep Learning, New IT Requirements for Edge Computing & MIPI Alliance for Mobile and IoT- Part II.

Introduction:

In this second of two articles on the JEDEC Mobile & IOT Forum we summarize three tutorials on I. AI/Deep Learning; II. Edge Computing requirements for the IoT and other latency sensitive or high bandwidth applications; and III. the MIPI Alliance for Mobile and IoT applications.  The first article can be read here.

I.  Making Sense of Artificial Intelligence – A Practical Guide:

This keynote presentation by Young Paik of Samsung was the clearest one I’ve ever heard on Artificial Intelligence (AI) – one of the most hyped and fudged technologies today.  Although it has existed in many forms for decades (this author took a grad course in AI in 1969), recent advances in Deep Learning (DL) and neural net processors have finally made it marketable. According to Young, there is real promise for AI and DL, but there are also real limitations. His talk provided an introductory overview of how AI and DL works today and some insights into different deployment scenarios.

DL has enabled AI to approach human level accuracy, as per this illustration:

 

A high level AI functional flow (but not implementation) and the Circle of Deep Learning (DL) Life are shown in the two graphics below.

In the second slide below, note that DL models need to be constantly fed data.  A home thermostat is used as an example:

………………………………………………………………………………………………………………………………………………………………………………………..

Mr. Paik’s said there are three takeaways from his talk:

1.  Data is King: The more data => greater the accuracy.

2.  Deep Learning is hard. Best to leave it to the professionals.

3.  You don’t have to use one AI: Many, smaller AIs are better than one big one.

…………………………………………………………………………………………………………………………………………………………………

The following illustration proposes functional blocks for implementing mobile speech recognition:

…………………………………………………………………………………………………………………………………………………………………………………………………

Two ways to improve DL are: Transfer Learning (take a pre-trained DL model and retrain it with new data) and Model Compressions (selectively remove weights and nodes which may not be important). Those “tricks” would permit you to remove several functional blocks in the previous illustration (above).

……………………………………………………………………………………………………………………………………………………………………………………………….

Finding new ways of using old tech and making use of multiple types of AI are shown in the following two figures:

…………………………………………………………………………………………………………………………………………………………………………………………………

Four different use cases or applications of AI are shown in this slide:

……………………………………………………………………………………………………………………………………………………………………………………………….

In conclusion Young suggested the following:

AI is still early in its development.

Design of AI systems is evolving.

You may find new uses for old ideas.

…………………………………………………………………………………………………………………………………………………………………………………………

II. A Distributed World – the New IT Requirements of Edge Computing:

The number of distributed, connected data sources throughout the world has been multiplying rapidly and are creating tremendous amounts of data. IoT has now given rise to a new trend of aggregating, computing, and leveraging data closer to where it is generated – at the IT “Edge” – between the Cloud and the IoT endpoint device. This presentation by Jonathan Hinkle of Lenovo provided insights into the new IT requirements for edge computing systems and models for how they are and will be used.

Jonathan asked: How do we leverage our IT resources to unlock the value of all the data now being generated?

Ideally, we should be able to gain many things from analyzing “big data” which includes: Business Insights, Optimize Services, Recognize Behaviors, and Identify Problems (when they occur).

IoT Architecture Components include:

  • Software: Analytics, Applications, Platforms
  • Security, Networking, Management
  • IoT Endpoint devices (“things”)
  • Edge IT (especially for low latency applications)
  • Core network and cloud IT

The functions required from the IoT endpoints to the cloud are: Generate Data /Automate / Control / Pre-process / Aggregate / Analyze / Store / Share the data. Observations:

  • It costs time, money, and power to move data.
  • Best practice: move data when useful or necessary
  • Reduce data set required to be forwarded to each stage

Keeping data local (instead of backhauling it to the cloud for processing/analysis) requires:

  • Store data nearer to sources (IoT endpoints) whenever possible. This is accomplished by filtering data at the edge such that less data (to be analyzed by powerful compute servers) are sent upstream to the cloud.
  • Maintain fast action on time-sensitive data by doing computation immediately instead of moving the data first.

In conclusion, Mr. Hinkle said that data growth will continue as the sources multiply – both from computing sources (e.g. smart phones, tablets, other handheld gadgets) and IoT endpoints which produce digital data to represent our analog world. “Edge IT infrastructure will enable us to scale with that data growth and unlock its inherent value.”

Author’s Note: Mr. Hinkle did not touch on how much, if any, AI/DL would be implemented in the “Edge IT infrastructure.” Unfortunately, the moderator didn’t permit time for questions like that one to be addressed.

……………………………………………………………………………………………………………………………………………………………………………………..

III.   MIPI Alliance – How Mobile Specifications are Driving IOT:

Peter Lefkin, Managing Director of MIPI Alliance, provided an overview of the MIPI Alliance and Specifications. Additional context, background and a a look ahead at the implementation of MIPI Specifications in mobile influenced industries such as IoT and automotive was also described.

MIPI Alliance specifications are developed for mobile devices as the primary target though MIPI Alliance members have leveraged and evolved their investments in mobile to other platforms including automotive and IoT.

For in IoT devices, MIPI specifications are implemented for various applications such as: augmented and virtual reality, wearables, and other low power and sensor connected devices.

Within automotive, automobiles have become a new platform for innovation and manufacturers are implementing MIPI specifications to develop applications for infotainment, advanced driver assistance systems (ADAS), and safety. Interconnected components for these applications include high-performance cameras and imaging sensors, infotainment and dashboard displays, telematics hubs among others.

Automotive and IoT platforms are heavily reliant on sensors and MIPI specifications will play a key role enabling sensor-based applications and connected devices.

 

Backgrounder:

MIPI is a global, collaborative organization founded in 2003 that comprises 312 member companies spanning the mobile and mobile ecosystems, including the Internet of Things IoT).

MIPI’s mission: To provide the hardware and software interface specifications device vendors need to create state-of-the-art, innovative mobile-connected devices while accelerating time-to-market and reducing costs.

In particular, the MIPI Alliance:

  • Defines and promotes specifications focusing on the mobile interface but applicable to IoT, Auto, etc.
  • Complements official standards bodies through collaboration.
  • Provides members with access to licenses as needed to implement and market specified technologies.

The MIPI membership list is at: https://mipi.org/membership/member-directory

MIPI Alliance for Mobile and IoT:

The MIPI Alliance serves the mobile industry and the ecosystem of mobile-influenced industries that are developing connected devices for vertical markets and the IoT..

MIPI specifications are crafted such that compliant devices have: high-bandwidth performance, low power consumption, and low electromagnetic interference (EMI). Here’s a MIPI Alliance mobile systems diagram:

https://www.mipi.org/sites/default/files/MIPI_SystemDiagram_Oct17..jpg

………………………………………………………………………………………………………………………………………………………………

Other Presentations:

Here are a few abstracts of presentations that may be of interest to IEEE techblog readers:

Comprehensive ARM Solutions for Innovative ML & AI Applications:

With the advent of AI and the explosion of ML/CV applications, there is greater demand for system solutions to enable vendors to get to market quickly. ARM is working on holistic system enablement, while allowing the flexibility and scope to incorporate additional hardware and software optimizations into customer platforms. This talk will discuss the work ARM has been doing in these areas to provide options for a stable and efficient software and hardware architecture.

Presenter: Ray Hwang, ARM

……………………………………………………………………………………………………………

Watts the Difference?

The challenge of designing to mobile and internet of things (IoT) markets is always the same question: How do you deliver a solution with the right combination of operating power, standby power, and environmental tolerance and yet provide memory capacity and performance? Hybrid combinations of DRAM, NAND, and NOR technologies have traditionally been used to balance these factors with varying degrees of success. Carbon nanotube memory (NRAM) allows designers to simplify the formula with a high capacity non-volatile memory device boasting DRAM-class performance and a high tolerance for heat.

Presenter: Bill Gervasi, Nantero

…………………………………………………………………………………………………………..

Flash Storage and Sensor Interfaces for Mobile and IoT SoCs:

Storage devices are a vital component of an electronic system for a variety of applications, and with today’s demand for faster boot-up and data transfer, choosing an optimized storage device has become a challenge for designers. For example, for mainstream smartphones, embedded technologies like eMMC has become the de facto storage device of choice due to its high-speed connectivity up to 400MB/s and cost-effectiveness. On the other hand, for high-end smartphones, UFS has become a robust option due to its unique high-performance, low-power and scalability advantages. Today, the use of such mobile storage devices extends into new applications like automotive and IoT. This presentation will describe each mobile storage specification and illustrate their unique use-cases and features such as command queuing, inline encryption and high bandwidth for mobile applications and beyond.

Presenter: Licinio Sousa, Synopsys

…………………………………………………………………………………………………….

 

4 thoughts on “JEDEC Forum: AI/Deep Learning, New IT Requirements for Edge Computing & MIPI Alliance for Mobile and IoT- Part II.

  1. Thanks for two terrific events summaries. Can you please provide more information on the MIP II Alliance which I have never heard of. Do they work with the IEEE Standards Board?

  2. Very interesting article! I expect the most popular use of AI will be at the network edge.

  3. Manufacturers like Qualcomm and MediaTek really emphasize utilizing cutting edge AI intelligence in their processors and chipsets. AI enhancements really up the gaming and over all performances of electronic devices according to these companies.

Comments are closed.