MEMS sensor fusion will promote rapid development of centralized processing

According to IHS iSuppli's Micro Electromechanical Systems (MEMS) and Sensors report, the trend of sensor fusion in the MEMS field is not only helping to improve the motion-based accuracy in mobile and gaming equipment, but also will bring new major issues to the field of centralized processing. Development, and is expected to substantially increase its operating income.

Currently ongoing developments in the area of ​​centralized processing are expected to use today's 9-axis sensor fusion, which uses a 3-axis accelerometer, 3-axis gyroscope, and 3-axis electronic compass simultaneously to provide more accurate and sensitive motion detection. Sales of motion sensors used in the 9-axis portfolio next year will reach US$8.50977 billion, a 55% increase from the US$5.5001 billion estimated at the end of 2011, and it is expected to grow to approximately US$1.35 billion by 2015, as shown in Figure 1. The incredible growth potential of sensor fusion may be beneficial to developers who focus on processing. They place their destiny on fusion technology.

The development of centralized processing means that the burden on application processors in smartphones, tablets, and game consoles will be unloaded elsewhere. Usually the application processor requires a lot of energy to continuously run various applications in the background. In addition, a mobile or gaming device requires a variety of sensors that require silicon to help detect a range of functions, including motion detection, ambient light, distance, temperature, and humidity. A large number of silicon chips are duplicated. As a result, the application processor is overloaded and consumes a large amount of energy. It is difficult to achieve the desired speed or achieve the maximum capacity.

One method of apportioning the application processor load is to use a microcontroller (MCU). Another approach, which is also likely to have a broader impact, will be to create a new class of hardware that is dedicated to sensor processing.

MCU

Compared to software algorithms that rely only on application processors, low-power MCUs can be used in conjunction with mobile application processors to reduce power consumption. In addition, vendors can add their own code to the MCU, giving it features that can help differentiate the product.

The main obstacle to adopting MCUs is cost. As an additional component, the MCU will inevitably increase the overall material cost of the cell phone or any device that uses the MCU.

So far, no mobile phone has used a dedicated MCU to undertake sensor signal processing tasks, but some mobile phone brands are working with MCU suppliers to develop related solutions. IHS believes that the first handsets with MCU sensors may be available early next year.

From a manufacturer's perspective, Texas Instruments is the first vendor to perform motion sensor processing tasks on its low-power MCUs. Another company based in San Francisco, Atmel Corp. , also announced in May 2011 with InvenSense Inc. of Sunnyvale, California. Cooperation. Their efforts are focused on consumer applications that require 6-axis sensor fusion, but Atmel may also develop solutions for mobile phones that use simple sensor fusion.

Other vendors are also working on the concept of dedicated MCUs for mobile phones. Freescale Semiconductor, Texas, USA, is promoting the concept of "smart sensor center", combining its low-power MCU and accelerometer in the same package; Italy-France joint venture STMicroelectronics, using an internally produced MCU Run its own 9-axis sensor fusion algorithm that provides signal processing for ambient light and distance sensors.

IHS Corporation believes that on the other side of the Pacific Ocean, Japan’s Roma Semiconductor is also studying how to use MCUs in sensor signal processing. Rom Semiconductor owns Kionix Inc. of New York. . Rom Semiconductor also provides a range of ambient light and distance sensors for mobile phones.

How is a dedicated hardware processor effective; MCU and dedicated hardware, which is better?

In a second solution to offload the application processor, the application processor vendor considers using a new type of specialized sensor processing core that can work while the application processor is sleeping. This dedicated kernel is useful for functions that require the sensor to continue to operate throughout the day, such as devices such as pedometers; it is also very effective for tasks involving action monitoring and situational awareness.

With a dedicated core, sensor fusion will be extremely rapid and effective. For example, calculating the coordinates will require fewer clock cycles. This hardware approach has many other advantages, including a significant drop in power consumption, which is very important for mobile devices; it eliminates the repetitive chip area involved in sensor signal processing. The vendor also does not need to develop special solution algorithms and can rely on the solution provided by this hardware in the application processor.

The use of dedicated sensors also has its disadvantages. In particular, application processor developers will need to invest large sums of money. While processor convergence is optimized for specific sensors—usually optimized for market leaders’ processors—the same accuracy and resolution levels may not be achievable for every sensor from every other vendor.

Chipmaker Qualcomm is clearly not aware of these potential challenges. It is believed that Qualcomm will soon introduce a new system-on-chip (SoC) with dedicated hardware for sensor fusion, an ARM 7 processor. The product will fully comply with the industry trend of reducing the burden on the application processor.

However, IHS believes that this step is a bit premature for the market. Although accelerometers and compasses in 9-axis fusion are mature technologies, 3-axis gyroscope technology is still improving. In particular, the new low-drift gyroscopes are not expected to be launched until 18 months later and may improve the performance of such products. Therefore, it is difficult to use the new gyro, which will be available in the near future, with hardware solutions that use dedicated sensors.

Overall, the current MCU approach is more feasible for reducing the burden on the application processor. On the other hand, after the gyro technology is mature and fully stable, dedicated hardware can be optimized to improve power consumption and response speed.

Ultimately, sensor fusion may have an additional impact on the motion sensor supply chain. Both MCUs and processor vendors can move up the value chain to a higher level. They can purchase sensor components or work with foundries to obtain dies. Integrate them with your own integrated circuit chip.