The Internet of things (IoT) aims to provide a more convenient and exciting way of life for consumers through the world around the Internet. But how do you achieve this commitment to the Internet of things?
In the world of Internet of things, MEMS sensor is a bridge between users and its surrounding devices, such as smart phones, wearable devices, robots and UAVs. However, only the sensing and connection of the equipment is not enough to achieve the big goal of the Internet of things. The Internet of things can be successful only if it is people-oriented, that is, to solve the daily challenges in real life, to make life more convenient and to improve ease of use. In addition, with the increasing complexity of the surrounding environment, the sensor suppliers are facing great challenges by all kinds of devices. These challenges and solutions will be explored from the perspective of the sensor vendors
Three challenges facing intelligent sensors in the field of Internet of things
Today's smart sensor modules contain some processing capabilities integrated with the original sensor, and the main challenges it faces can be attributed to the following three points:
The first challenge is technology itself. The supplier wants to use its core MEMS and system technology to accomplish this impossible task. For engineers, this is a challenge to physical restrictions. The size of the package can not be reduced indefinitely, and the requirements for low energy consumption and high performance are also increasing. The supplier has to improve the system to make it smarter and more intellectually aware. To achieve this goal, technology must be made across multiple product platforms.
The second challenge stems from the wide spread of the industry. At present, most of the benefits of MEMS sensors come from smart phones, which sell more than one billion smartphones every year, and each smartphone contains at least one MEMS sensor. According to the specifications set by the smartphone original equipment manufacturer (OEM), Bosch Sensortec and other manufacturers have developed a corresponding MEMS sensor.
But the Internet of things is a special field, which is characterized by the high decentralization of the competitive technology platform structure. In the entire Internet of things space, the requirements for sensor subsystems consisting of sensors, microcontrollers and actuators are very different. As a result, Bosch Sensortec and other vendors need to create cross platform solutions for integrated hardware and software and provide dedicated applications. With software and professional application technology, vendors do not have to customize hardware solutions for each application when helping customers to solve specific problems.
The final challenge is the complexity of geometric growth. The Internet of things system itself is very complex. Only providing components can't meet the needs of original equipment manufacturers. Usually, one-stop solution or reference design is needed. The market leading supplier will incorporate more system processing capability into a single modular device, and develop a solution for integrating smart sensors based on this, so as to meet the demand of greatly reducing complexity. Because no company can provide a comprehensive solution, suppliers must cooperate closely with the third parties in creating reference designs, and establish partnerships.
The hierarchical structure of the sensor information of the Internet of things
The information structure of the Internet of things includes several levels. The hierarchy of typical applications is sorted by the increasing status of information usefulness as follows:
Sensor information hierarchy
1. original data
2. motion detection
3. activity monitoring
4. situational perception
5. intention prediction
Although the original data may be filtered, supplemented, and changed, in most cases they clearly limit the user's use of data. At the second level, the data are analyzed by the recognition mode and the application algorithm to provide the motion detection information. Then, by adding additional sensor functions, for example, according to the height of air pressure measurement, we can enter the next level to allocate information for inference activity monitoring. In today's ubiquitous computing environment, the definition of device context awareness is more abundant: interaction with other devices, adaptation to environmental noise and light conditions, and network status. This makes the task more complex: for example, a weighted assessment based on situation and behavior patterns is needed to generate predictive decision-making.
At this point, the way the sensor system processes data is comparable to the human brain function. The human brain mainly uses two systems in data processing: the cognitive system and the edge system. The cognitive system is similar to cloud computing - high physical efficiency and large capacity memory with delay. By contrast, the edge system is a system of original, reactive and reflective, which corresponds to the local processing in the sensor system, that is, the edge calculation.
Figure 1: describe the difference between cognition and limbic system, analogous to human brain's processing of perceptual data (image source: [Krisdog] / Depositphotos.com; Bosch Sensortec).
In the Internet of things, the level of sensor information is important to determine whether the measurement data is important or not. Do not have the possibility of basic data can not be used, resulting in a large number of miscellaneous redundant data exist in most applications.
Intuitively, it is easy to keep the sensor permanently and wait for the identification of useful information, such as the accelerometer in the step application, which is usually more effective. The sensor system needs to be intelligently screened for data to be transmitted to the cloud, so that available bandwidth and power are used effectively. The key is that the local sensor can automatically filter out most of the useless data, thus saving the valuable system drive capacity.
Internet of things system driver
In the application of the IOT sensor, there are several key system drivers that affect the design of the system and components:
Low energy consumption is of vital importance to some applications of small or portable devices. In this case, the autonomous sensor processor matching the sensing element (sensor robot) helps edge processing, that is, when to transmit data to the cloud, so as to reduce the resource cost of data transmission.
It is particularly important to shorten the delay time in the case that a large amount of data is transmitted in the shortest time. For example, in virtual reality (VR), the image needs to be sent in real time so as to synchronize with the motion of the user's head.
In the system of fast behavior learning, high data sampling rate is essential. For example, in the prediction and maintenance of a vibrating machine, the sensor must be sampled at a high enough rate to capture all the related data that leads to the failure of the device.
The variables of simple integration are very large. Due to the length of input and the size of engineering resources, the expectations of original equipment manufacturers to sensor data are often different. In order to simplify the integration of sensors in applications, more and more companies use smart sensors to match data processing devices embedded in sensors with software solutions provided by suppliers. For example, in the field of robotics, the original equipment manufacturer is more focused on the motion of the robot itself and tends to completely do not deal with the original sensor data.
The edge operation is similar to the above edge system. We sometimes need edge processing, which is usually a precondition for low power and easy integration.
Because of the very high cost of memory in the sensor modules, cloud storage becomes a viable alternative for local storage and processing. On the one hand, we do not want to transmit a large amount of unnecessary data, on the other hand, we are limited by the physical storage capacity of the sensor. Therefore, we have to intelligentize the sensor and ensure that the sensor can filter out most of the useless data and prevent out of memory.
Let's give an example to illustrate the points of the picture. First of all, wearable applications, such as pedometer must never power and the battery must be as small as possible. The key to this kind of application is low energy consumption, which is realized by integrating the step function directly into the sensor itself. Unless absolutely necessary, it will not wake up the main processor of the wearable device and save battery power.
In order to save resources, users can not transfer all the data to the host, which is also a typical feature of the edge computing application. From the energy consumption point of view, the ultra low energy consumption solution of BHA250 or BHI160 is an ideal choice.
Another example is the development trend of rapid prototyping technology, which is increasingly widely used in market verification by large companies. Rapid prototyping technology is usually used in development platforms such as Arduino, Raspberry Pi or other open source systems, including the combination verification of sensor components.
This kind of application requires sensor vendors to provide relatively complex software to achieve the maximum integration. The development time must be as short as possible, and the original supplier also needs to use limited sensor knowledge to explore the design of the system. The application of sensors to multiple platforms, such as Arduino and Raspberry Pi, can greatly simplify integration.
To achieve the success of the Internet of things applications, a qualified sensor provider that understands this highly complex environment of the Internet of things is needed as a partner. They are able to provide a wide range of high performance sensor combinations and provide high quality solutions for customers' applications. Quality, local support, and strong third party partners are equally important, which can provide reference design and system level expertise.
The Internet of things requires a deep understanding of many applications and meets the requirements of various sensors and processing, such as low power consumption, easy integration, data rate and short delay. Only by understanding the relationship between these different factors can we design innovative high-quality products for the fast developing IOT market and make users' life more convenient, so as to achieve the promise of the Internet of things.