Let’s explore the technology inside IoT endpoints
The Internet of Things (IoT) is often measured in endpoints. An endpoint is any connected device that provides raw or processed data to the network. But IoT is bigger than the billions of smart devices sitting at its edge.
In the last two decades the IoT has created new ways to connect old equipment. More importantly, it has inspired semiconductor manufacturers to develop new technologies.
An IoT endpoint may be in service for years or even decades and sometimes without so much as a battery change. This puts emphasis on reliability and energy efficiency. New technologies must recognize these imperatives. Engineers need to choose wisely when developing new features using the latest integrated devices.
Machine learning in the IoT
If your IoT application isn’t already using artificial intelligence (AI) or machine learning (ML), you are undoubtedly considering it. The IoT is alive with AI. This intelligence is moving closer to the edge and, in some cases, right to the very edge. This relieves the burden of sending large amounts of raw data over what is often a limited connection. It also removes the latency incurred with processing data in the cloud.
AI and ML are being used to detect patterns. Using AI to detect events that do not fit a pattern can be even more useful. These anomalies may indicate a fault when used for predictive maintenance. Image recognition is also moving closer to the edge, to provide pre-qualification of objects or activities. Gesture recognition is another edge-based application that fits in endpoints, making use of AI.
Engineers may also be evaluating the best way to add intelligence to endpoints. New microelectromechanical systems (MEMS) enabled sensor technology from STMicroelectronics integrates the company’s intelligent sensor processing unit (ISPU). This highly efficient processing unit executes AI algorithms on the sensor itself, so it needs no additional processing resources from a host microcontroller. It can be used to detect movement and identify types of motion, including gestures. It uses the data coming from the integrated three-axis accelerometer and three-axis gyroscope.
Microchip Technology developed a kit based on the SAMD21 microcontroller for evaluating applications that use AI at the edge. The kit features an IMU from TDK InvenSense.
A new family of microcontrollers from NXP, the MCX, integrates a neural processing unit (NPU). This is a dedicated computation core that executes ML algorithms on the device. NXP says it can result in ML that runs 30 times faster than on a standard microcontroller core.
Using ML to detect patterns in vibration and motion sensor data is becoming a key application in the IoT. Evaluation kits such as the SAMD21 Machine Learning Evaluation Kit with TDK IMU from Microchip Technology provide an easy route to getting started.
The SAMD21 kit comprises the SAM-IoT WG Development Board and a MikroE 6DOF IMU 14 Click board. The sensor used is the ICM-42688-P high precision six-axis MEMS motion tracking device from TDK InvenSense. The microcontroller on the main board is a SAM-D21G18 and it also includes an ATECC608A CryptoAuthentication secure element integrated circuit (IC), along with an ATWINC1510 Wi-Fi network controller.
Security in the IoT
The open nature of the IoT means security breaches are not uncommon. This isn’t a failing of the underlying infrastructure, but rather a historical lack of prioritization for security features. OEMs now have access to a growing range of preconfigured and easily integrated security solutions.
The OPTIGA Trust M from Infineon is a turnkey solution to securing an IoT endpoint. The device integrates tamper-proof, non-volatile memory to securely store information, with both asymmetric and symmetric crypto engines that support commonly used security measures, such as elliptic curve cryptography and the Advanced Encryption Standard (AES).
Signing and verifying are important stages in establishing a secure connection between an endpoint and the cloud platform. Managing secured devices in the field is another factor that OEMs should consider. Secure device management (SDM) is a service from Avnet’s IoT solutions team and can be used for onboarding and authorizing devices, managing OTA updates while in service, and the secure offboarding of devices at the end of the working life.
IoT connectivity
Connectivity is the obvious essential in IoT. Wired and wireless communications carry the data that defines IoT. The network’s core uses internet protocols to provide compatibility throughout the network. At the endpoint, the options are not restricted to internet protocols. Gateways provide the conduit between endpoints and the core. This increases the options for developers.
Wireless communication is a clear favorite in the IoT. It provides a simple, cable-free connection for endpoints. Some common wireless technologies include Bluetooth, Wi-Fi, Zigbee and Z-Wave. These have traditionally been used for short distances (or personal and local area networking). However, the need for longer range has been met through developments that extend their reach.
Longer wireless range generally comes at the cost of more power. This is where the low-power wide-area wireless protocols can offer an advantage. The compromise is normally in bandwidth, but with intelligence moving to the edge, many IoT applications don’t need to send and receive large amounts of data. Leading low-power wide-area network (LPWAN) technologies in this space include LoRa, Sigfox and, through cellular networking, CAT M1 and NB-IoT.
Pre-certified wireless modules can provide a fast and simple way to add connectivity to an IoT endpoint
The STM32WLx5xx series of wireless microcontrollers integrates an ultra-low-power LPWAN radio subsystem. This means they comply with the physical layer requirements of LoRaWAN, a long-range wireless protocol. LPWANs like LoRa, Sigfox and M-Bus are becoming more popular for low bandwidth IoT applications such as smart environmental sensors and meter reading.
Some wireless protocols are IP-based, including Wi-Fi, Thread and 6LoWPAN, but one of the newest IP-based communication technologies impacting the IoT isn’t wireless. Single Pair Ethernet (SPE) is gaining interest in the IoT thanks to its lighter weight and lower cost. It can also carry power over the same conductors, which makes it convenient for smart sensors and other endpoints.
The appeal of SPE and IP-based wireless protocols is part of a general effort toward creating a seamless link from sensor to cloud. Using IP-based technologies makes each endpoint accessible through its own unique IP address.
Using wireless technologies for location and proximity measurement
As well as providing a way of exchanging data, wireless technologies can deliver location services using indoor positioning. We know GPS doesn’t work well inside buildings. Other wireless technologies can address this. Several wireless protocols are adding support for indoor positioning, including Bluetooth, Wi-Fi and Ultra-Wideband (UWB).
Indoor positioning systems based on wireless protocols use beacons in a fixed and known position. The beacons communicate with assets in the wireless area. As well as data, the RF signal coming from the assets can provide enough information for the beacons to calculate their position. Parameters measured include the strength of the signal at the beacon, the angle of arrival and angle of departure of RF signals (which can require multiple aerials), and the time of arrival.
The fixed beacons use these parameters to calculate the distance to the movable asset and from that its relative position. Knowing the fixed position of the beacons allows the system to calculate the absolute position of the asset.
While an absolute position can be useful, a relative proximity measurement can also be helpful in an IoT application. One of the newest methodologies for distance measurement in IoT applications uses optical time of flight (ToF). A ToF measurement comprises a light source and a receiver. The light source is tightly coupled to the receiver. Distance is calculated by measuring the time taken for the emitted light to be received when reflected by an object in its field of view. Light patterns are used to provide accuracy and immunity from stray sources of light in the area.
The ISL29501 from Renesas is a ToF signal processing IC for low power and long-range optical distance sensing. The IC drives an external LED or laser, and a photodiode for detecting the reflected light. It can operate in continuous or single-shot modes and has on-chip active ambient light rejection.
Conclusion
Endpoints in the IoT are evolving. It is no longer just about adding connectivity. The value of actionable data is increasingly becoming critical to business success. With new technologies coming to market, Avnet is working with its supplier partners to bring the latest and most innovative solutions to its customers.
Contact us to find out more about Avnet’s IoT solutions and how your business could benefit.
Avnet works with leading semiconductor, electromechanical and software suppliers to put the latest IoT technologies in the hands of product developers.
LEARN MORE BY VISITING OUR DEDICATED IoT MICROSITE.