Watching Processor IP Trends in the Smart Wearable Market
As shipments of wearable devices continue to rise, smartwatches have evolved to be one of the most popular devices in this sector. To make the most of a projected 42.8% compound annual growth rate (CAGR) developers are examining what qualities will be needed to drive adoption.
A report from analyst firm IDC, forecast that worldwide wearable device shipments will exceed 200 million (214.6 million) units by 2019. Smartwatches are the more popular type of wearable, with a projected 34.3 million units to be shipped this year, rising to a forecast of 88.3 million units by 2019, a five year CAGR of 42.8%.
According to Ramon Llamas, Research Manager, IDC Wearables: “Cellular connectivity, health sensors, not to mention the explosive third-party application market . . . will raise both the appeal and value of the market.”
Kinjal Dave, Senior Product Manager, ARM®, believes that complexity, power consumption and price point are significant factors in smartwatch design evolution.
Figure 1: The smartwatch market is the largest sector in wearable devices, and is forecast to reach over 88 million units shipped by 2019.
Both ARM Cortex®-A and Cortex-M processors are used in nearly all of the market leading smart devices, including the Samsung Galaxy Gear S2 and the Motorola Moto 360, which both use the ARM Cortex-A, and the Pebble and FitBit, which use the ARM Cortex-M.
The difference between these two processor series, explains Dave, is in the computational requirements they have been developed to address. The Cortex-A line-up, ARM’s application processor series, is generally targeted for use cases where rich Operating Systems (OS) are required, running a variety of software applications, often from multiple sources. Cortex-M cores, on the other hand, have been designed to address the ever-expanding microcontroller market for smaller, energy sipping and deeply embedded, real time applications.
Cortex-M cores are designed to have very low interrupt latencies, he says. This addresses the growing embedded market, as sensors or motors connected to a Cortex-M processor can be controlled quickly in real-time. Cortex-M devices are generally clocked lower and have simpler architectures. This combined with their lower gate count can lead to devices that use a tiny amount of power.
The Cortex-A series is best suited to applications where a rich user experience is required. For example, where the screen or user interface and OS are used to interact and also end-devices that require more bandwidth, more data and are expected to process that increased data quickly, for example a mobile phone or tablet. Cortex-A devices tend to have higher clock rates and more complex designs to maximize computational performance.
Whilst these are the general characteristics, there are equally many cases where Cortex-A is used in more traditional embedded applications. In addition there are Cortex-M devices that offer significant performance. The wearable device market is an interesting example of where both classes of processor have gained traction.
Delving into more technical detail, one of most significant differences between Cortex-A and Cortex-M cores is the memory architectures. Cortex-M processors have simpler memory architectures, to suit the needs of applications that generally do not have as much data, but require consistent deterministic response.
The complexity of applications that require user interface interaction, large amounts of data processing and bandwidth demands, and generally more complex software, drives a requirement for a more complex memory architecture. The Cortex-A’s Memory Management Unit (MMU) addresses this, allowing the use of virtual memory. Whilst a full explanation of virtual memory is beyond the scope of this article, this allows the processor to run significantly more complex programs requiring a much larger address space.
Figure 2: The Cortex-A32 is the latest Cortex-A processor, introduced at Embedded World, in February.
While the Cortex-A processor is ideal for the rich, graphical smartwatch interfaces associated with Android Wear and similar platforms, a Cortex-M based device pushes the boundaries of what can be done with a more cost-sensitive microcontroller core. Cortex-M processors are generally ultra-low-power in operation, explains Dave. The Pebble interface is an example of how a microcontroller can be used to provide an effective graphical interface, while maximizing battery. The wide spectrum of capabilities on offer from these two architectures have enabled product developers to design to a variety of trade-off points between rich user experience and capabilities, and battery life and cost.
Another trend worth noting, says Dave, is the use of Cortex-M processors for capabilities that require a small amount of processing capability, but require extreme efficiency for always-on operation. A good example can been seen in activity monitors such as in the FitBit. Cortex-M allows always-on sensing in these devices, whilst maintaining days of battery life. Increasingly, a Cortex-M based sensor-hub is seen alongside a Cortex-A processor, offloading this always-on processing from the application processor, while allowing the support of a rich operating system and user experience when the user actively uses the device. The application core can ‘sleep’ whilst the Cortex-M handles sensor processing, and only wakes when the user requires it.
Figure 3: Cortex-M processors are efficient to prolong battery life in wearable devices which require always-on operation.
Software and OS
The smartwatch market is dominated by the Apple Watch and by Google’s Android Wear. Other popular OS include Linux, Pebble OS, and RTOS-based watches (IDC Worldwide Quarterly Wearable Device Tracker, March 2016).
While Cortex-M based devices tend to use a Real-Time Operating System (RTOS), those built around Cortex-A cores tend to use Linux-based environments. Most wearable Linux-based operating systems have been derived from mobile platforms, and leverage years of progress and development in this market. These offer third party developers a good platform on which to develop apps, often offering tight integration with established mobile app ecosystems.
Linux based operating systems also provide the rich UI toolkits to provide sophisticated graphical interfaces to users. An RTOS running on top of a Cortex-M can still provide an engaging experience, but there is a trade off in terms of complexity and capability, and app ecosystems are not as common. Pebble is an excellent counterpoint to this trend, however, developing a great user experience and an expanding app ecosystem around a Cortex-M based device.
The different focus of the Cortex-M and Cortex-A offerings allows for developers to adopt a different focus and differentiate end-products. This, in turn, provides more choice for consumers. While wearable devices such as smartwatches or fitness trackers make an excellent case study, this breadth of innovation can also apply to embedded, industrial, IoT and automotive projects.
This article was sponsored by ARM.