Many applications need to process and refine their data at the edge.
Data management at the edge cannot be overlooked because this will be instrumental in driving efficiency and providing real-time system insight through analytics, thereby helping embedded systems to make smart decisions.
Ignoring the need to modernize and manage data where it originates, on microcontrollers (MCUs) at the edge, will be a huge mistake. According to various reports, the exponentially increasing number of connected devices will create a data tsunami, and smart embedded systems must act as wave breakers for the ocean of data.
As part of this, MCUs will need flexible data management software that enables sensor data to be dynamically processed, stored, and communicated. This will enable each MCU to take on new responsibilities as part of the connected ecosystem.
When you build an application with MCUs, important considerations include memory footprint, total cost of ownership (TCO), and component integration. Generally, each MCU project uses a unique combination of hardware components that greatly influences software requirements. A major challenge for MCU applications is to process and refine the data at the edge with limited resources, so only relevant useful information is collected, organized, and stored. An effective solution will minimize footprint, data processing overhead, and long-term data maintenance costs.
Software components are typically deployed to an MCU as a unified firmware image. By design, the firmware is too infrequently updated to keep up with changes to the way sensor data should be analyzed and processed. Embedding a modern MCU database into the firmware provides the framework to organize data, run dynamically created data collection campaigns, and control device peripherals.
Unlocking the full potential of time series data pouring in from sensors into MCUs at the edge will require the right database. For a long time, applications built with MCUs performed in an isolated environment. However, with the arrival of the connected world and the internet of things (IoT) revolution, this has changed. Building and embedding a database into firmware is not a simple operation. Great data management expertise and know-how is required to develop, test, and debug a data storage solution. The arrival of machine learning and artificial intelligence for MCU applications will also shift expectations. Preparing and performing all these tasks is not trivial and carries a significant cost.
Over the years, we have worked with many customers building applications with MCUs. Some customers require data storage for only simple data types; some are building complex edge systems for which data processing, security, and integration with other components play a vital role; and others want to accomplish much data management and storage at the edge, but also to send data to the cloud.
We have seen manufacturers building various applications that collect log data and update configuration data to perform preventive maintenance, anomaly detection, and more. These embedded systems gain intelligence, make sense of data, and provide real-time insight. We’ve designed and developed our database for the IoT—ITTIA DB IoT—for MCUs organically, according to what we have learned.
Our main objective has always been modernizing data management for collecting, smoothing, and organizing real-time data at the edge. Our data management offers the filtering, aggregating, and smoothing of useful information, while also extracting relevant data from sensors and IoT nodes living at the very edge of the internet.
When you decide to consider a database for your application, it is important to make sure that you select a robust and verified database solution that has been designed from the ground up for tomorrow. Why? Because as you grow your business and develop additional requirements for integration and data exchange, obsolete solutions will only work with a portion of your data, not all of it. Data is a new currency. No one likes to go to the bank and have access to only a percentage of their funds.
Another important factor is security. There are many MCU application developers that purposely ignore adding security features to their products. Why? Because security adds memory footprint and they are unwilling or unable to allocate extra memory.
An additional essential issue is the type of data for which your application needs support. For us, time series data are the main citizens, so this is how we architected ITTIA DB. A time series database is a software system that is optimized for storing and serving time series through associated pairs of time(s) and value(s). The time series database is purpose-built to handle metrics, events, configuration, and logs.
When working on building platforms to manage MCU data, developers frequently ignore the importance of data management. Most systems built with MCUs will be deployed for a long time and—as the embedded systems grow—their data management complexities rise. It is important to select the right database vendor familiar with related challenges with the right database software for MCUs. The experienced database software vendor provides involvement and know-how for microcontrollers and offers a comprehensive review of data processing and management.
Processing and storing data closer to where it’s generated and eliminating the need to send it all over the embedded system or to a centralized network offers great bandwidth and data maintenance cost savings. It also offers increased performance as data can be processed instantly. OEMs demand edge data management software that allows them to process and analyze data at the device so their system can gain insight and take immediate action. Even analytics and device training can be implemented on the MCUs.
Performing data management at the MCU edge allows these OEMs to have a great number of IoT devices working together, improving data management performance, and offering customers great cost savings.
You are building an embedded system that should last for ten, fifteen, or even twenty years (or more). We always encourage our customers not to underestimate the knowledge, cost, and complexity of the integration of MCU applications with databases. We believe one of your first tasks should be to determine the complexity of your data management and then select an appropriate database solution that will increase the power, performance, and longevity of your product.