How will both software and hardware components need to expand to meet the demands of IoT and smart home automation?
That is a great question. Personally, even though I’ve been working on IoT at Avnet for a couple of years now, I am not an early adopter myself of smart home automation. Why? Because I want Nirvana, and I suspect most customers will as well.
Where are we today? People like to use words like “fragmented” and “diverse” to describe it. Many companies are competing for the ‘hub’ of the system, and there are competing wireless standards as well such as WiFi, Bluetooth Mesh, Z-wave, Thread, and ZigBee to name a few. This amounts to a tangled web of standards, software, and vendors. To reach the Nirvana I described above, I would probably need three or four hubs to manage each major system, each of which would need to communicate with my router, and none of this implements a locally autonomous system as a whole.
Can we get to Nirvana? It will happen. On the software side, there is good news. We are already seeing many vendors integrate with the major automation cloud services like Apple Homekit, Amazon Alexa, and Google Assistant. However, we need to see coalescence of wireless standards. Once we have one, or at most two, accepted wireless standards, in addition to WiFi, then the major consumer router manufacturers will jump into the fray with products that can communicate and control most connected devices, driving the price down.
Finally, the home automation gateways (as I think routers will eventually become) will have to be intelligent in their own right. There needs to be local (in-house) processing to be able to handle video security tasks; enact automation rules, such as what to do if a user flicks a light switch; and so on. This will happen soon, and then my home automation will have reached Nirvana.
By David West, Director of Professional Services, Icon Labs
IoT devices are predominantly price-sensitive and deployed outside of a secure perimeter with a very long life cycle. In most cases, cost, more than any other factor, drives security component selection.
When choosing between hardware or software, the best solution is to build security into the device and not depend upon the perimeter. Typically, on-device security is an order-of-magnitude lower cost. Addressing basic security needs like an embedded firewall and secure boot cost-effectively protect the device from both inside and outside attacks.
Likely candidates for hardware solutions include Physically Unclonable Functions (PUF), Trusted Platform Module (TPM), and TrustZone.
PUF uses random patterns in the silicon to differentiate chips from each other and to create a unique random number. The generated random number is used to seed a strong device ID and cryptographic keys creating a hardware root of trust.
Security co-processors are physically separate chips offering true isolation of private keys. A TPM offers isolation along with crypto functionality, key generation, and secure storage. However, its cost usually moves it to higher end IoT devices.
Trust Zone is another single chip solution that segregates execution space into secure and insecure worlds. Unsecure apps can’t access security-critical assets. Those same security critical assets are isolated from tampering. Like a TPM, cost moves it to higher end devices.
Software security provides a layer of protection at a much lower cost while offering a broader range of options compared to hardware. Frequent candidates for software security include a firewall blocking unwanted packets, TLS/SSH for secure communication, intrusion detection, and management functions. Compared to hardware solutions, software may consume more power.
Ultimately, some combination of hardware and software will be required. Only the system designer will be able to make that determination based upon costs and likely attack vectors.
By Cristian Ionescu-Catrina, Sr. Marketing Manager, Home Appliances
The cloud with its huge amount of resources brings the promise to provide almost endless processing and storage, however not for free. There are also situations when the connectivity link may not reliably work, depriving connectivity dependent smart devices of their “smart” appeal.
Today’s smart hub refrigerator concept represents a good example for connectivity dependency with at least three embedded cameras always sending pictures to the cloud, where also the vision processing software is running. With huge amount of data /device sent to the cloud and full cloud dependency for image processing such solution is difficult to scale to millions of devices both from cost and operability perspective.
We believe that a hybrid approach will be able to address above challenges. By using a hybrid solution, the vision processing task is partitioned into two pieces: embedded-AI on the edge, which is focusing on decisions the software has been trained for, and the cloud, where the machine learning/improvement resides. In the case of the fridge, the embedded-AI can identify items which are stored on a regular basis and are already a part of the library (For example eggs, milk, or soda cans). In this case, cloud connectivity and processing is not needed.
However, as soon as there is a new food item/brand in the refrigerator which the embedded-AI can’t recognize, the edge will provide the images to the cloud which in turn will try to classify the data. As a result, the total cost of ownership goes down and the edge increases its operability. We envision future MCU and MPU will integrate dedicated hardware IP blocks which will be a (partial) hardcoded implementation of popular artificial intelligence software, like voice or image.
By Mark Tekippe, Senior Marketing Manager, IoT Wireless Products, Silicon Labs
The smart home is one of the fastest growing IoT segments with billions of new connected devices expected between now and 2025. Several analyst reports estimate the average smart home will include from 50 to 100+ connected devices including appliances, lighting, security systems, and a myriad of wireless sensors. This rapid growth will only be possible through advancements in hardware and software components to address the following key needs: simple installation, seamless interoperability, uncompromised security, device management for field updates, and ultra-low power.
Simple installation and seamless interoperability of IoT products across different manufacturers and wireless protocols are essential. The advent of multiprotocol, multiband wireless SoCs will enable device manufacturers to support multiple connectivity interfaces like WiFi, Bluetooth, and 802.15.4 (ZigBee/Thread) in their products while keeping device hardware costs in check. For example, connected products will be able to leverage Bluetooth for simple setup via a smart phone and 802.15.4 for reliable, scalable mesh networking with other devices—without significantly impacting device cost. In addition to supporting common RF protocols, true interoperability requires devices to speak a “common language,” which requires standardization at the application layer (i.e. dotdot, Weave, and OCF). Multiple “languages” will co-exist in the home, and automation software platforms will need to manage protocol translation.
Uncompromised security demands continuous innovation in both hardware and software. Advanced hardware cryptography, integrated secure elements, secure firmware updates, physical tamper detection, and end-to-end (device-to-cloud) IP security are some of the key features required to secure the connected home.
Device management is imperative to provide reliable and secure mechanisms to update firmware on devices long after they are initially sold and deployed in a home. Similar to how mobile app developers frequently push updates to their apps, connected product manufacturers need the ability to enhance product functionality and address bugs and security vulnerabilities without requiring a truck roll or product return.
Finally, nobody wants to replace batteries on hundreds of products in their home. Improvements in battery technology, energy harvesting, and low-power operation of wireless SoCs and sensors will extend battery life and enable large-scale networks in the home.
The future of home automation is bright, but also highly dependent on hardware and software innovation to bring home automation to the mainstream and realize the smart home vision.
By Bill Steinike, Vice President of Business Development, Laird Connectivity
Consumers have high expectations for these products and are savvy buyers because of the time they invest researching products. I see four key criteria that engineers should focus on to meet those rising expectations.
Security is critical, but can’t be at the expense of making the commissioning difficult or laborious. In some products, enhanced security has come at the cost of a more cumbersome implementation process for customers. But a straight-forward commissioning approach should not be readily sacrificed to achieve improved security; after all, it is the first experience the user will have with the product and plays such a big role in overall customer experience. Security and ease-of-use don’t need to be in opposition to one another, though. The most successful products will be the ones that offer both intuitive, scalable commissioning, and security features that can respond to evolving threats.
Another key factor in a product’s success is wireless performance as reliable and robust as wired installations. Consumers no longer grade wireless products on a curve. They simply must perform as well as any other device. Luckily, advancements in WiFi with features like MU-MIMO and the upcoming 802.11ax standards, Bluetooth 5.0 and its inclusion of longer range and higher data-rate capabilities, and the variety of options for LPWAN technologies like LoRa will better enable performance along with required interoperability.
Another area where customers are very savvy is battery life. It’s a key buying criteria for the consumer electronics they purchase, and the same is true for home automation products. Hardware must achieve better battery efficiency, especially for relatively power-hungry technologies like WiFi that have a broad appeal to many developers. Customers won’t tolerate the hassle or expense of frequent battery changes.
The last thing that I will mention is real-time locationing. Improvements in the performance and cost to make indoor locationing capabilities more widely adopted could vastly improve the user experience for many dynamic IoT applications, particularly in building and home automation.