Overcome Smart Home Technology Limits Using Sensors and Edge Computing
Designing smart home devices involves numerous challenges. In this article, learn the important limitations of today’s smart home technologies and how sensor fusion helps smooth the way.
A smart home system can include devices related to security, kitchen appliances, lighting, entertainment, and heating/cooling. As the technology for smart homes has advanced, engineers must develop more capabilities and more intuitive, intelligent products. And these expectations include high levels of accuracy, faster performance, and easier integration of multiple devices (Figure 1).
Figure 1. Modern intelligent home technology includes widely varying intelligent devices and massive amounts of data that needs to be processed quickly. Image provided courtesy of Pixabay.
There are limitations to improving smart home technology, but they can be addressed through contextual data provided by a combination of sensors and processed on the device rather than in the cloud.
Smart Home Technology Limitations
As with any technology, the fundamental components and systems are constantly improving. Engineers are tasked with continuously developing better solutions as soon as limitations are recognized. Three major limitations in smart home technology are accuracy, latency, and compatibility.
Accuracy Affects Everything
Accuracy is extremely important in smart home technology and is affected by everything from the sensors that collect data to the AI (artificial intelligence) tools that process that data. Engineers are taking innovative new approaches to collect data, such as combining multiple sensors and using algorithms to process the data together to achieve higher levels of accuracy.
For example, smart home security systems may combine computer vision, radar, and sound detection to notice when someone is present accurately. New algorithms and AI tools are also being developed to find the most efficient ways to process data. This, however, leads to another challenge: latency.
Latency is critical in any type of smart home technology, but especially in home security, where data from multiple sensors must be combined and analyzed as quickly as possible. Increasing the amount of data gathered, transmitted, and processed will significantly impact latency.
If the system takes too long to process the data, it does not matter how accurate the results may be if an alert or decision is issued too late. Engineers must find ways to reduce the latency present in smart home devices and systems.
The ecosystems in which smart home technology must function also impact performance and functionality. End users may have multiple smart systems working in their homes (for example, Amazon Alexa and Google Home) as well as devices from different manufacturers.
Engineers are moving their focus away from ecosystems that are specific to platforms, manufacturers, and devices. Instead, they are moving more of the processing and functionality to the devices themselves. And there is one design approach that can address all three challenges: edge computing.
Edge Computing—Camera Example
In the context of smart home technology, edge computing shifts the majority of data processing and analysis to the device itself rather than to the cloud. A more technical definition is that the data is processed as close to the source as possible.
Let’s examine the example of home security cameras. As many of us know from personal experience, these cameras often report false positives, and an abundance of these errors can eventually cause us to ignore accurate alerts.
A common approach to improving the accuracy of home security cameras often focuses on the quality of lens and image sensors to more accurately differentiate between people and other objects that may appear to be moving. While this is a reliable method, there is a better approach.
Improving camera performance through better lenses and more accurate (and expensive) image sensors is fine, but they don’t provide any contextual data to help the system make more intuitive decisions.
Although those camera and sensor improvements can enhance a system's ability to detect motion accurately, context is needed to differentiate between leaves moving in the wind or an animal passing by.
Figure 2. Radar sensors such as the one shown can be used to develop intuitive home security devices that provide contextual data. Image used courtesy of Infineon
Context can be provided through advanced microphones that can detect the sound of footsteps or the crash of a breaking window or radar that can provide rich information on the nature of the motion, allowing the system to tell the difference between moving tree branches and an approaching human while providing critical data on the direction of motion.
Here’s a summary of common sensor options for smarter home security cameras:
- Radar detection sensors: should be size-optimized with AIP (antennas in package), such as the one shown in Figure 2. They should also include built-in motion detectors and the ability to determine the direction of motion, all taking place on the chip.
- Analog microphones: high precision, high SNRs (signal to noise ratios), for example greater than 70 dB, that are IP-rated weather
- MEMS barometric pressure sensors: should combine sensing and conversion into an electrical signal on a single chip
Adding these additional sensors means significantly more real-time data is generated that must be transmitted and processed before the security system knows if an alert needs to be generated. This can be addressed through sensor fusion and edge-based computing.
Sensor Fusion Brings it All Together
Sensor fusion combines the data from multiple sources to support context and provides the AI machine learning tools needed to analyze the fused data accurately. In this approach, all fusion and processing take place on the smart device (in this case, a smart home security camera) rather than in the cloud.
This eliminates the time involved in data transmission and avoids issues related to bandwidth and connections outside of the local network. The result is increased accuracy and reduced latency. However, a low-power microcontroller (MCU) is necessary to achieve sensor fusion and processing at the edge, such as an Arm Cortex M4F class or better.
This approach also addresses an issue with the ecosystem in use because processing and alert generation takes place at the device level. This allows it to be easily configured by engineers for compatibility with a wide range of systems.
Infineon Solutions for Smart Home Security
Infineon offers an extensive line of smart, edge-based home security systems that combine data from multiple systems to maximize accuracy and minimize false positives through contextual awareness. Their system solution includes sensors, software for sensor fusion and machine learning/AI, MCUs, and developer boards.
Infineon XENSIV sensors for contextual, edge-based smart home devices include MEMS analog microphones with a sealed design for ingress protection, high SNF, and high precision, all in a very compact package shown in Figure 3. Both 24 and 60-GHz radar sensors can also be found in the XENSIV line along with pressure, temperature, and CO2 detection sensors.
Figure 3. The XENSIV IM73A135 MEMS microphone offers high performance and a wide dynamic range. Image used courtesy of Infineon
To support edge-computing and low latency, Infineon provides their SAS Fusion software and machine learning tools with training capabilities, including SAS Fusion software. There is the PSoC IoT designed Arm Cortex M4F class or better MCU with embedded flash and excellent power efficiency for processing.
To facilitate the design of edge computing smart home products, Infineon also has sensor kits, design boards, and online simulation software to facilitate the design of edge-computing smart home products.
Getting it Right for Smart Homes and the Edge
There are many applications for more intelligent, intuitive edge-based smart home devices. These include service robotics, office devices, and white goods with onboard intelligence.
Infineon offers highly accurate, effective, and system-level solutions. Edge-based solutions engineers can use those solutions to create applications such as automotive, communication, industrial security, and consumer goods. Importantly, Infineon also has what’s needed for contextual, low-latency, ecosystem-compatible products.
Industry Articles are a form of content that allows industry partners to share useful news, messages, and technology with All About Circuits readers in a way editorial content is not well suited to. All Industry Articles are subject to strict editorial guidelines with the intention of offering readers useful news, technical expertise, or stories. The viewpoints and opinions expressed in Industry Articles are those of the partner and not necessarily those of All About Circuits or its writers.