The wearable devices currently in the market seem to fall into two categories: they either come with long battery life but simple features, or short battery life with comprehensive features. Is there a solution to make wearable devices that have both long battery life and comprehensive features? Imagination has the answer.
No compromise: Wearable devices are not mobile devices
In his speech, "Get that smartphone chip out of my wearable! Designing SoCs optimized for wearable computing," Bryce Johnstone, Head of Ecosystems at Imagination Technologies, stated that currently, wearable devices such as smartwatches mostly adopt smartphone chips resulting in products that are high in power consumption with unnecessary features. To increase sales, chips for wearables should be redesigned. Johnstone first talked about power consumption and safety of wearable technology and then introduced new chip solutions designed by Imagination's partners specifically for wearable devices.
Johnstone noted that wearable devices have been in the market for many years, such as the heart rate monitor introduced by Polar in the 1980's. With the coming together of a number of technology and social aspects, the time for wearables has now arrived. Advances such as configurable and flexible processing, low-cost silicon, low-power connectivity and others are combining to meet emerging requirements today's highly tech savvy consumers, who want instant social media interaction, personal and health monitoring, and always-on connectivity. The industry has begun focusing on wearables, and is providing low-cost and low-power consumption MEMS sensors, advanced power management technology, and chipsets designed specifically for this market.
Mobile and wearable devices have very different hardware structures
According to Johnstone, the wearable devices currently in the market usually adopt compromised solutions.
The main similarity between today's smartwatches is they tend to adopt high-end chips and hardware similar to those used in smartphones. This means the products consume power more rapidly and users have to recharge frequently. Also the user interface of these products is similar to that of smartphones, but varies among brands because there has not been a standard application programming interface (API).
Wearable devices should not just be a smaller version of mobile devices, but should instead meet demand from different markets and provide specific and necessary, often complementary functions. In designing these products, companies must consider the many factors, including: low power consumption (power consumption should be calculated on a daily or weekly basis for exercise sensors, monthly for home care sensors, and yearly for vital sign sensors), low cost, small memory, method of use, data security, data accuracy, data collection efficiency, data management, and connectivity (low frequency bandwidth and high efficiency power use and data transfer, support low power-consumption mesh network, and wireless connection).
Johnstone outlined some of the necessary hardware specifications for smart bracelets, smart watches and smartphones for comparison. In terms of the CPU speeds, an example smart bracelet runs on a sub-100MHz MCU, compared to an example smartwatch with a 0.3-1GHz multi-core power-saving application processor (AP), and a smartphone with a 2GHz quad-core CPU.
As for other items in the comparison, display screens are unnecessary for smart bracelets. Smartwatches need only a small screen with a simple interface. Smartphones obviously feature larger displays that support a full range of functions. As for memory and battery capacity, it is low for smart bracelets, medium for smartwatches and high for smartphones. However, battery life for smart bracelets is long, medium for smartwatches and short for smartphones.
Different wearable devices require different functionality. For example, smart bracelets need various types of sensors to accurately collect information; smart watches provide a more convenient way to read messages and should focus on simple interactions; and smart glasses provide high resolution images and camera functionality and focus on expanding the virtual reality experience.
One potential wearable device architecture outlined by Johnstone features a 3-CPU core architecture: a wearable computing CPU, wearable control CPU, and sensor CPU (which connects various sensors).
Johnstone also showed the difference in power consumption between wearable devices with smartphone components and wearable devices with wearable-optimized components. Smartwatches with wearable components can be 10 times more efficient in power consumption performing exercise monitoring functions (with only the sensor CPU on), 10-20 times more efficient in voice recognition and operating various apps (with the wearable control CPU on).They can three times more efficient in running an OS like Android Wear, which was introduced by Google to specifically accommodate wearable devices with an aim to provide the best user experience and a standard API (with the wearable computing CPU on). This means the multiple-CPU architecture meets the requirements of wearable devices.
Power-saving and high efficiency is the best wearable device solution
Imagination's partner Ingenic has introduced Newton, a MIPS-based wearable hardware development platform. The Newton platform is embedded with low-power consumption and high-performance JZ4775 1GHz MIPS CPU, and can process multimedia (enabling 2D graphic engines and various types hardware acceleration). It features multiple memory support (256KB L2 cache RAM + up to 3GB mobile DDR3/DDR2/LPDDR RAM + eMMC), contains a four-in-one connectivity component (Wi-Fi a/b/g/n, 2.4/5GHz, Bluetooth 4.0, EDR supporting Bluetooth, NFC, and FM), and supports various sensors (habitual, environmental, and biological). The device consumes power around four microwatts during standby, and 100 microwatts during operation. The PCB size is the size of two one-euro coins.
Another Imagination partner, Ineda, has introduced the Dhanush WPU (wearable processing unit), which is the first processing unit designed specifically for the wearable device market. The system structure incorporates the three types of CPUs as described above: a low-power consumption sensor CPU, a low-power consumption sub CPU, and a main CPU for application processing. In particular, the sensor CPU only operates Tiny RTOS to collect data, hence only consumes one-third of the power compared to pure MCU solutions. The sensor CPU consumes 2-3 times less power during standby, supports 2-30 times more memory, and the embedded DSP/FPU runs 2-3 times faster than pure MCU solutions.
Johnstone concluded that MIPS structure is the CPU with the most potential for expansion in the wearables market. It supports products ranging from low- to high-performance applications. It is an ideal choice for various sectors.
Bryce Johnstone, Head of Ecosystems at Imagination Technologies