Ekkono
Making the Smart Devices Smarter

By giving the inanimate physical world a digital nervous system, the Internet of Things (IoT) has been a hot topic for quite a while now. However, this technology in its current state is far from perfect, the biggest shortcoming being the equipment’s need for manual input to obtain optimal function. For instance, if a person wakes up at 8:00 A.M. on any typical day and wants his/her coffee exactly 15 minutes later, they can encode the smart coffee maker to brew at 8:15 A.M. daily. But this will be an optimal setting only if the person follows his/her morning schedule to the highest order of strictness. Since such a discipline cannot be exhibited by everyone, on aggregate, this hard-encoding will not exhibit optimal functionality. Against this backdrop, incorporation of machine learning can make the smart devices smarter by analysing the data generated by the connected equipment to get an insight into the human’s behavioural pattern. As a result, the devices can feel a little bit more like an assistant and a lot less like a liability that has to be encoded.

To that end, Ekkono, a Sweden-based IoT company, provides a highly configurable, small-footprint, platformagnostic, embedded software library that is built for the purpose of helping developers rapidly and easily deploy Edge Machine Learning. This technology enables the developers to learn the normal state and behaviour of their connected things, which is the baseline to detect anomalies, attribute what has the biggest impact on performance, predict coming events, and even simulate what if-scenarios.

As the word means ‘cognition’, Ekkono licenses embedded software and the tools to make connected things smart. Not just using machine learning for one feature, but virtually making all functions self-aware and predictive, the company empowers product developers to rapidly implement edge intelligence themselves using its programmable tools. Ekkono’s toolkit facilitates use cases like predictive/condition-based maintenance, self-configuring products, performance optimisation, and entirely new business models. “We are in a perfect storm where more and more people in the industry have understood the great advantages of combining technology for edge computing, machine learning, and automation,” says Jon Lindén,Co-Founder and CEO of Ekkono. “Right now, many people are crazy about us and want to start piloting our technology.”

With an all-software and totally platform-agnostic product, Ekkono helps harmonise on one solution for every kind of edge machine learning needs. The company’s embedded edge machine learning library, purpose-built for IoT comes with an SDK (software development kit), which is a comprehensive toolbox to support implementation and integration. These tools help with algorithm selection and optimisation, enabling even the programmers with limited data science experience to deploy advanced edge machine learning features. This product supports the execution of pretrained models, as well as incremental learning at the edge, where the data pipeline is automated and built into the process. It also supports a number of machine-learning techniques, including decision trees, random forest, and neural networks. Besides incremental learning, Ekkono also offers unique features such as conformal predictions, efficient techniques for anomaly and change detection, and more.

Digging deeper, Ekkono’s uniqueness also lies in its ability to do incremental learning on streaming sensor data onboard the device. This stems out from its seven years of machine-learning research at the University of Borås. Ekkono allows the processing of all the high-frequency sensor data in real-time with a layered approach. This not only results in the ability to take instant actions but also to learn and act based on the normal condition of the individual device. Ekkono radically reduces network load by not sending raw data, but rather sending relevant and enriched data to the cloud for further analysis. By making the connected things become autonomously smart and less dependent on the quality of the network connection, the company has pioneered the art of making devices more intuitive.