Years ago, sensors were essentially data monitors in standalone machines or wired networks. The rapid growth of the IoT has thrust sensors into a pivotal role in monitoring and communicating data to locations that could be only a few feet away or up in the cloud. This rich data is enabling businesses and institutions to run more efficiently and optimize cost and available resources.
But that’s only part of the story. Sensors are teaming with innovative semiconductors and leveraging developments in disruptive technologies such as AI (artificial intelligence) and machine learning to solve challenges in areas ranging from home security to medicine to industrial and manufacturing applications. The recently completed Sensors Converge Conference and Exhibition in San Jose showed several examples of these technologies converging into innovative solutions.
One example is home security. Infineon Technologies showed a battery-powered Smart Alarm System system that achieves high accuracy and low-power operation using sensor fusion based on AI and machine learning. The system incorporates Infineon’s analog XENSIV™ MEMS microphone IM73A135V01 , XENSIV digital pressure sensor DPS310, and PSoC™ 62 microcontroller. To make this solution work, developed a sensor fusion software algorithm, based on AI/ML, that combines acoustic and pressure sensor data to accurately differentiate between sharp sounds inside a home and distinctive audio/pressure events.
“What we found was that existing alarm systems are often prone to false alarms,” said Dave Jones, Head of Marketing & Business Development, IoT & Sensor Solutions America at Infineon Technologies, in an interview with Design News. “Lots of sounds correlate to glass breaking, fireworks, or dog barking that are false alarms. Glass breaking is the most difficult case as its sound resembles others in the audio spectrum.”
Jones explained that the AI-developed sensor fusion algorithms modeled patterns rather than the sound signatures in frequency-based sensor models, enabling the alarm to weed out events that could generate false alarms.
Infineon is offering the SAS in a reference design designed for use in WiFi networks.
Long Battery Life
But Infineon is not alone. Aspinity, which designs and develops analog machine learning chips, announced at Sensors Converge a glass break sensing technology that also eliminates the false alarms common with household or other common soundsꟷwhile promising battery life lasting up to five years.
Aspinity’s solution revolves around its AML 100 analog machine learning chip, which leverages the company’s RAMP (Reconfigurable Analog Modular Processor) neural processor algorithms to eliminate the digitization of irrelevant data and its associated wasted power. This system-level approach eliminates the power penalty of digitization, digital processing, and transmission of irrelevant data, effectively reducing always-on system power by >95% to a mere 125µW (AML100 and a microphone) for the 99.9+% of the time that there’s no glass. The chip enables an array of security and home automation applications requiring an extended battery lifetime, including the detection of T3/T4 alarm tones (smoke and carbon monoxide, respectively).
Not surprisingly, sensors used in industrial applications are also benefitting from technology innovations. Analog Devices unveiled a high-resolution depth sensing module, designated the time-of-flight sensor, the ADTF3175, for 3D depth sensing and vision systems used in industrial applications. The time-of-flight sensor uses IP from Microsoft and achieves a resolution of 1 megapixel, which Analog Devices claims is not matched by currently available time-of-flight sensors.
“The applications for the module are anywhere requiring a realistic 3D user experience,” said Erik Barnes, Product Line Director of Time of Flight Sensors at Analog Devices,” in an interview with Design News. Robots, AR/VR, consumer applications are some of areas where high-resolution 3D is becoming important, according to Barnes.
Designed for easy integration into end-user applications, the module incorporates an infrared illumination source with optics, laser diode and driver, and a receiver path with a lens and an optical band-pass filter. The module also includes flash memory for calibration and firmware storage plus power regulators to generate local supply voltages. The pre-programmed module comes with several operating modes optimized for long- and short-range.
Sensors are playing an increasingly important role in medical applications, and these use-cases were also on display at Sensors Converge. ams OSRAM introduced its AS705X series of vital sign sensors. These biosignal converting devices combine a highly customized optical front end in conjunction with a power-saving design and small form factor for incorporation into hearables, smartwatches, or smart patches.
The first of these sensors, the AS7050, capture biosignals from heart rate measurement (HRM) and galvanic skin resistance (GSR). Thanks to two ADC (Analog to Digital Converter) channels and embedded ECG low noise analog front end, the AS7050 can perform photoplethysmogram (PPG) and electrocardiogram (ECG) measurements simultaneously. Customers who require an even smaller optical set-up and have no need for an ECG can opt for the AS7056 or the AS7057 family members.
Wireless IoT Sensing
Powercast Corp., a supplier of radio-frequency (RF)-based over-the-air wireless power technology, and InPlay, inventor of the programming-free, ultra-low-cost and low-power Bluetooth sensor NanoBeacon system-on-chip (SoC) technology, demonstrated a platform to design battery-free, maintenance-free, long-range wireless IoT sensor systems for deployment in retail, medical, warehousing and industrial IoT markets.
Powercast's far-field wireless technology can now power InPlay’s NanoBeacon IN100 120 feet away from the commercially-available PowerSpot® transmitter, enabling a beacon signal every minute. NanoBeacon's low-power design helps Powercast achieves this performance. As the NanoBeacon moves farther from Powercast's RF transmitter, beacon signals become less frequent, while moving it closer allows more frequency if the application requires it.