Early 2018, and it’s still an exciting road trip for the IoT industry. Connecting those 8.3 billion smart devices and sensors of all varieties is indeed making the Internet wiser. But will the 3X growth rate from 2015 to 2020 in connected things – up to 21 billion connections, per that well-quoted Gartner report – mean that the Internet will also become three times smarter? Depends on how well those devices perform.
Growing IQ across the IoT network
Simple monitoring has been a decent use case for early IoT deployments, confirming that data stream conversion and IP connectivity can generate a cloud’s worth of big data. Now the focus is on pushing more compute and storage out toward the edge of the network, closer to the devices themselves. Artificial intelligence and machine learning, for example, are occurring across the compute continuum, riding on new smart devices. From Gartner’s 2018 predictions: “As the technology develops, AI and machine learning will increasingly appear in a variety of objects ranging from smart healthcare equipment to autonomous harvesting robots for farms.” Moving your workloads from the cloud or core to the network edge helps bring down the cost of analytics in the cloud and improve the security, latency, privacy and bandwidth of the IoT network.
Industrial-quality memory devices
The Industrial IoT (IIoT, or Industrie 4.0 in Europe) especially benefits from devices with local AI and continuous, sensor-driven analytics. IIoT applications are often tuned to real-time action and algorithms to control devices at the edge to overcome the problems of cloud latency, which can be detrimental. IIoT hardware devices become the conduit to selling services, providing a way to monetize data and analytics, so benefit from rigor in the design. The focus is often on making them cheaper and faster, but it’s better to think about full life-cycle cost of the product and delivering a lower total cost of ownership (TCO) to end customers rather than just the initial bill of material (BOM) of the device.
Longevity is also important, and managing obsolescence costs should be considered in the TCO. IIoT deployments take time, and devices can reside in remote or hazardous environments where replacements are a challenge. You’ll usually want to identify components with long-term availability from vendors with roadmap stability and extended product life cycles.
IQ—Industrial Quotient—is the mindset to make intelligent choices in your industrial product design that help you meet not only the functional product requirements, but ensure quality and reliability of your product, long term, in a wide range of use cases, without compromising the product life cycle management. IIoT and other workloads vary by use case and context, but incorporating high-endurance industrial-grade memory and electronics can make a real difference in performance, consistency and access.
Flexible memory and storage for device-IQ use cases
Communication: Multichip packages (MCPs), which combine two technologies into one package, accommodate the tight space constraints and memory requirements of communication modules. Connectivity and communication is fundamental to IoT deployment. Most industrial and commercial systems deployed in wide areas require cellular connectivity for communication back to aggregation gateways or to the cloud. Most cellular communication modules (2.5G/3G/LTE) require specific MCP solutions due to space and cost constraints.
Smart Transportation/Fleet Management: For better asset tracking and fleet scheduling to platooning, placing sensors inside the vehicles is an efficient approach. These systems are generally designed using industrial computer on module or industrial PC solutions that require DRAM, NOR flash, and managed NAND. While e.MMC devices are typically used for code, application software and data storage, low power DDR or LPDDR4 devices can support the bandwidth demands of new high-performance analytics applications. Some use cases also need ruggedized industrial SSDs if the data capture requirements exceed embedded storage capacity or a removable storage solution is required by the application.
Security/Surveillance: The adoption of edge storage and increased edge device intelligence in cameras, plus the massive improvements in image sensors, resolution and digitalization, have progressed professional-grade surveillance. For example, with 64-layer 3D NAND technology, it’s now possible to build highly reliable video surveillance microSD cards with ultra-high densities, like 128GB and 256GB. This opens a future of hybrid deployments where edge storage is used in conjunction with traditional NVRs, which can significantly reduce bandwidth, storage and maintenance costs and increase flexibility with cloud-based deployments.
Cyber-security at device endpoints
Cyber-security continues to be a major concern with distributed system architectures like the IoT which requires a different mindset than how we design devices for managed enterprise networks. Look to flash memory, one of the most common components in all IoT devices. As the repository of critical device code, flash memory is also the place to add security and ruggedize the system against firmware-level attacks. By combining the unique device-specific identity only a hardware root of trust can offer, along with the measurement capabilities necessary for in-memory secure boot, solutions like Micron Authenta provide a strong cryptographic fingerprint to authenticate IoT devices directly with a host, such as a secure gateway, or from an end-point to the cloud. Integrating this with the IoT cloud solutions like Microsoft Azure IoT suite or application-specific networks like NetFoundry MultiCloud Connect solution helps create an end-to-end secure IoT solution.
As IoT devices at the network edge become smarter and more secure, organizations can leverage their Industrial Quotient to transform industry and business.