Leading Wearables into a New Era with Cutting-Edge Connected Bluetooth LE based MCUs

author avatar
Leading Wearables into a New Era with Cutting-Edge Connected Bluetooth LE based MCUs

Learn how bluetooth connected AI/ML microcontrollers are setting the bar for next generation wearables.

Wearable electronic devices designed to be worn while in use are becoming more popular and powerful as they can monitor various aspects of human health, activity, and environment. However, these devices also face challenges such as limited battery life, data privacy, and network bandwidth. To overcome these challenges, edge computing and AI/ML are emerging as key technologies that can enable wearable devices to process data locally, reduce latency, and enhance user experience. 

As wearable technologies evolve to meet market needs, new devices are upgrading features such as basic BLE connectivity by adding the ability to run AI algorithms on neural processors on-device rather than in the cloud. This reduces the amount of data that needs to be transmitted over the network, saving energy and bandwidth. With the AI/ML models running in silicon, wearable devices can, therefore, learn from data and perform tasks such as recognition, classification, prediction, and optimization without incurring additional latency, data costs, and power consumption. However, these improvements require greater processing power and capabilities from embedded microcontrollers (MCUs) and application processors. At the same time, consumers expect longer battery life from their wearables. Thus, new chips must balance connectivity and higher processing power with low power consumption while maintaining a small enough package size. 

This case study explores the challenges for the next generations of wearable MCUs with a specific look at Alif Semiconductor's® Balletto™ family of BLE-enabled MCUs for wearable applications.

The current landscape

The wearables market, which includes common accessories such as smart bands, smart rings, smartwatches, and medical devices, such as glucose monitoring devices is projected to grow to USD 118.16 billion by 2028 at a 13.8% CAGR. Demand for multimedia devices, smartphones, fitness trackers, and health wearables drives growth. Increasing health awareness fuels interest in fitness tracking devices for activities like cycling and running, contributing to market expansion.[2,3]

Wearables have evolved significantly in recent years, presenting enhanced capabilities in various domains, such as AI processing and wirelessly transmitting data. New market demands have driven the expansion of wearable features. For example, the COVID-19 pandemic made blood oxygen sensing a must-have feature for many smartwatches. As a result, Apple and Samsung included blood oxygen monitoring in their smartwatches in 2020.

Wearables incorporate edge AI-based solutions to process huge amounts of data in real-time. AI-enabled microcontrollers are a must to carry out these large AI/ML workloads efficiently and quickly. The integration of artificial intelligence enables new features, such as:

  • Tracking and analyzing changes in users’ health data.

  • Sending users alerts for noise levels or air quality.

  • Personalize your device based on your tracked activity. 

Equally important as collecting and processing user data is the ability to wirelessly share data between devices. As such, Bluetooth connectivity has risen as a top requirement for wearables including functionalities, such as: 

  • Sending an alert to emergency services if a fall is detected.

  • Sharing important health updates with an approved health care professional.

  • Sharing workout metrics with a friend to stay motivated.

Bluetooth LE changes the status quo in wearables

Among the many low-power wireless technologies on the market, Bluetooth Low Energy (BLE) technology is uniquely suited for wearable devices that require long battery life measured in days or even weeks before needing to be recharged. BLE allows wearables to easily connect and communicate with other devices, such as smartphones, tablets, exercise equipment, or even other wearables. 

This enhances the device's functionality by enabling it to share exercise data, receive timely notifications, and more. Increasingly, wearable devices are adding audio support for which BLE is uniquely positioned. With the addition of support for the Low Complexity Communication Codec (LC3), BLE is now capable of use cases such as broadcast audio and shared or group listening experiences with LE Audio’s Auracast feature. For example, a wearable device can use the on-device neural processor such as Arm’s Ethos-U55 to run a convolutional neural network (CNN) model for gesture recognition or activity classification based on the motion data captured by inertial measurement units (IMUs). The device can then use BLE to communicate with other devices or a smartphone app to provide feedback or instructions to the user. 

In the Smart Home, wearables can be integrated into a smart home system, with AI algorithms learning the user's habits and preferences to automate various home appliances. BLE 5.4's enhanced performance and range can facilitate efficient communication with many IoT devices in the home. For example, a wearable could learn that a user likes to listen to calming music when they get home from work and automatically instruct the smart speaker to play the user's favorite playlist when they walk through the door. 

Additionally, features allow wearables to be located precisely, while location-based services (LBS) such as personalized fitness coaching (created by learning from a user’s previous workout data) can be enabled. It can suggest workout routines tailored to the user's fitness goals and current health status. The wearable device can recognize the user’s activity and posture using an ML model that runs on the Ethos-U55 core and sends the data to a cloud service via Enhanced Attribute Protocol (EAP). This could enable personalized, reliable, and secure fitness tracking for the user. 

Next-gen wearables will enable highly-personal fitness tracking. Credit: WHOOP.

 AI/ML algorithms can be trained to detect patterns in biometric data gathered by wearables, such as heart rate, blood pressure, and oxygen saturation levels. BLE's enhanced broadcast communication mechanism (EAD, PAwR) can allow these devices to send real-time health risk alerts to other devices, such as a smartphone, even in congested or high-interference environments. For example, a wearable could alert a user or a healthcare provider if it detects signs of an impending health crisis like a heart attack or stroke. 

When you combine AI-enabled MCUs with BLE, wearables become more personalized, efficient, secure, and useful in our everyday lives. Alif Semiconductor has just announced their Balletto family of BLE and Matter microcontrollers with neural processing for AI/ML workloads, which seeks to exceed the standard for low-power BLE operations.

Hardware accelerated machine learning

To meet the market-need for state-of-the-art wearables, Alif Semiconductor offers AI/ML-enabled microcontrollers with dedicated NPU accelerators that allow heavy AI/ML workloads to run without compromising battery life. The Alif Balletto family of microcontrollers introduces the world's most power-efficient microcontroller with on-device secure AI/ML processing, top-notch security, and ample peripherals for applications requiring integrated Bluetooth Low Energy and 802.15.4-based connectivity. 

The Balletto family boasts the powerful Arm® Cortex®-M55 core, equipped with double-precision FPU and Helium Vector Extensions, operating at a remarkable speed of up to 160MHz. Arm Ethos-U55 is a dedicated NPU accelerator that offloads AI/ML operations from the Cortex-M55 processor, allowing the M55 cores to operate in very low-power modes. This is especially important for wearables like hearing aids, which need to have a long battery life. Equally as important for wearables is having a small form-factor design, which Balletto can provide in a sub 16mm ² package.   

Balletto’s Ethos™-U55 Neural Processing Unit (NPU) can improve the performance and efficiency of speech recognition machine learning models by up to two orders of magnitude, excluding the 5-10x improvement already provided by the Cortex-M55 core over previous generation Cortex-M CPUs. This is ideal for wearables that rely on voice commands, improving processing speed and lowering latency. 

Balletto and Ensemble families. Image credit: Alif Semiconductor.

Battery-friendly microcontrollers powered by Alif’s aiPM™ technology 

The Balletto family also tackles a common problem seen by previously existing AI/ML-powered microcontrollers: short battery life. Connected wearables or mission-critical devices like hearing aids depend on long battery life for users. Their ability to function is hindered when there is a need for heavy local processing and AI/ML. Balletto MCUs are the most power-efficient in their class, thanks to Alif's innovative aiPM technology. This technology manages all aspects of the device, including individual processing cores for real-time control, neural processing, DSP and audio, deep security, and radio protocols capable of BLE 5.4 with LE Audio, Thread, and Matter. Balletto devices also include a wide range of interfaces and large memories, all monolithically integrated inside extremely small packages. This makes them ideal for demanding battery-powered applications such as wearables and hearables.

Alif Semiconductor’s aiPM technology manages the power usage of the device through 8 power domains. It activates only the necessary chip sections and deactivates them when not in use, resulting in a highly efficient always-on region when there is a high requirement for local processing and AI/ML. The integrated dual-PA and on-chip antenna gives customers control of the reliability-range equation by providing tuning options of 4 dBm or up to 10 dBm Bluetooth transmit power.  Balletto’s Bluetooth radio has a receive current of 1.5mA and a transmit current of 2.0mA, making it power conscious.

Additionally, off-die peripherals produce higher energy consumption for wearables, while Balletto’s on-die solution fosters maximum efficiency. Utilization of the internal bus fabric effectively lowers access time, dynamic power, and standby power, ultimately providing more granular and precise power control at a lower cost. 

Balletto family multi-layered security

In addition to handling processing requirements, microcontrollers for wearables must ensure data security. Alif Balletto microcontrollers provide exceptional security features with multiple layers of protection, including:

  • Root of Trust

  • Unique device ID for each individual chip

  • Dedicated security processor

  • Dedicated protected memory

  • Configurable firewalls that regulate access of each CPU to sections of memories and individual peripherals and extend the capabilities of the standard Arm Trust Zone security partitioning

Alif Semiconductor is the only company in this market segment that injects at manufacturing, a unique public key infrastructure (PKI) key pair to support immutable hardware root of trust (RoT) from secure boot to strong authentication and remote software attestation capabilities.

As a result, the personal data acquired by sensors and processed by the Alif Balletto microcontrollers are secured against third-party attacks.

Users can securely manage all aspects of the device lifecycle, including key management, certificate management, remote updates, and more. Legacy MCUs may stop at TrustZone, but Alif takes you beyond TrustZone to ensure that multi-core devices are protected by not allowing two CPUs to access the same Secure domain and share secrets. 

Graphics and imaging

The Balletto family includes a number of hardware features to help offload traditionally CPU-intensive workloads from the MCU/MPU cores to improve performance and keep current consumption in check. One such feature is an integrated 2D graphics accelerator that can help speed up drawing operations by 4-10x compared to an unaccelerated system. This, combined with a set of imaging and display interfaces, which include parallel as well as MIPI-CSI2 and MIPI-DSI interfaces for 24-bit RGB displays and 8 to 16-bit cameras, provide everything needed for responsive and buttery-smooth graphics on pixel-dense screens.

Conclusion

As the wearable technology market continues to expand, the convergence of edge AI and BLE technology has enabled wearables to become smarter, more efficient, and more secure. Alif Semiconductor's Balletto family of MCUs represents a significant leap forward in meeting the ever-growing demands of the wearables market. These innovative solutions empower wearables to not only collect and process data but also to provide real-time, personalized, and secure experiences, all while maintaining long-lasting battery life and optimizing form factor. The future of wearables is here, and it's exciting.

References

  1. GlobeNewswire. 2023- 2028 Smart Wearables Market Size, Share and Trends Analysis Report By Y-O-Y.

  2. businesswire. Global Wearable Technology Market Trends & Analysis Report 2021-2028: Adoption of Fitness Trackers and Health-based Wearables is Anticipated to Propel Growth - ResearchAndMarkets.com.

Further Research

Ramesh Nair, 2023. “Smart Wearables: Advancing Safety in the Workplace

StartUs Insights, 2023. “Top 10 Wearables Trends in 2023

GlobalData, 2022. “Wearable technology market continues to rise

Lisa Eadicicco, 2022. “Smartwatches Have Measured Blood Oxygen for Years. But Is This Useful?

Fi Forest, 2022. “Wearables: managing complexities in data privacy and consent in healthcare tech

Industry News, 2022. “Alif Semiconductor partners with Telit to develop and deploy IoT edge devices

Alif Semiconductor, 2023. “Supplying next generation scalable microcontrollers

Arrow, 2022. “Alif Semiconductor

Eduardo Montanez, 2022. “NXP i.MX RT MCU Technology Powers Our Smartwatch Future