The word ekkono means cognition, and that is what we add to connected things. Ekkono makes machine learning for IoT. This enables you to learn the normal state and behavior of your 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. This is commonly used for proactive maintenance, operational optimization, and to make your products more intuitive.
The uniqueness, and the result of 7 years of machine learning research at the University of Borås in Sweden, is the ability to do this at the edge of the network, on the actual device. The benefits are significant. You can process all the data in real-time. More sensors equals more data, and the only way to manage this flood is through a layered approach that starts by the data source. You can take instant actions. You learn and act based on the normal condition of the individual device. You radically reduce network load by not sending raw data, but rather sending relevant and curated data to the cloud for further analysis. Your connected things become autonomously smart and less dependent on the quality of the network connection.
IDC: ”By 2019, 40% of digital transformation initiatives – and 100% of IoT initiatives – will be supported by AI capabilities.”
We provide the solution for this 100%. IoT and machine learning are tightly connected. Manual supervision of billions of connected things will ruin the tremendous business prospect that IoT represents. Instead the machines have to become a little smarter. We, Ekkono, make them smart. This translates into money as you avoid unscheduled downtime, reduces the need for manual supervision, optimizes production and performance, makes customers happier, and makes better use of available network and processing resources.
IoT Solutions World Congress (IOT SWC), Fira Gran Via, Barcelona, Oct 16-18, 2018
Ekkono exhibits at Europe's leading IoT event that takes place in Barcelona in October. Have a look at the conference program where Ekkono's CEO, Jon Lindén, is one of the speakers/panelists. For more information, go to https://www.iotsworldcongress.com.
MobilityXlab selects Ekkono: "MobilityXlab welcomes new emerging companies"
MobilityXlab, co-owned by Volvo Cars, CEVT, Volvo Group, Ericsson, Veoneer (Autoliv) and Zenuity, has selected and invited Ekkono as one of six companies for the Fall term of 2018. "We're seeing extensive interest from the automotive industry, and MobilityXlab is a fast-path into projects with the innovation leaders of this space", says Ekkono's CEO, Jon Lindén.
The industry news site and portal, AI Business, did an interview with Ekkono's CEO, Jon Lindén, in preparation for the AI Summit in London in June. Read about challenges and experiences of implementing AI in Industrial IoT.
Stacey On IoT Interviews Ekkono's CEO On Innovative Use Cases
IoT blogger, podcaster and influencer Stacey Higginbotham, aka Stacey On IoT, quotes Ekkono's CEO, Jon Lindén, on innovative use cases that don't capitalize on selling user data; "Data is not the new oil, but it could keep customers loyal".
Ekkono Quoted As Participant in CEVT's InnoScout Initiative
Ekkono's CEO, Jon Lindén, was quoted in CEVT's press release about InnoScout after his presentation yesterday. CEVT, China Euro Vehicle Technology AB, is a Geely company (Volvo Cars etc.) working with the next generation of cars. InnoScout is an initiative from CEVT to engage and interact with innovative startups, like Ekkono. Read the press release here.
Ekkono In The News – "Walerud Ventures Invests In The IoT Company Ekkono" (Di Digital)
Sweden's largest business daily wrote an article on Ekkono's new investors. Read the article here (in Swedish).
IoT (Internet of Things) holds great promises of everything getting smarter – from homes, cars, robots, and vending machines, to cities. But this requires the things to get smarter, which is done using machine learning. Smarter things is a necessity, or the great prospect of IoT will be lost in a tsunami of sensor data and business cases that are broken by the labor cost for manual supervision of this data flood.
The only way to manage this flood is through a layered approach that starts all the way out on the device, close to the data source. Ekkono applies machine learning on all your sensor data, in real-time – data that cannot be sent to the cloud due its sheer volume. An airplane generates 2.5 TB per day, an autonomous car 10 GB/mile. Not even 5G is the solution for this, which is why many IoT deployments today rely on blunt averages of historical data that are uploaded to the big data haystack. Ekkono feeds curated and relevant data to the cloud for further analysis and cross-referencing across the entire installed base. It also enables instant actions when events occur, and you can handle the data in a more delicate way by sending hash-like KPIs instead of raw data.
This is how we take IoT from connected to smart – and even autonomously smart. This means that it works even when the connection doesn't. Many connected things move around, or are in geographical or physical locations – like a forest or a basement – with challenged connectivity. Edge smarts works anyway. And it gets personal. Ekkono learns the behavior of the specific device, and provides tailored responses. Machine learning enables multiple variables to be combined into complex correlation models, which for example means that the pressure indication of a machine can still be normal, due to heat, humidity, current production rate etcetera, even though it is above the general threshold for its type of machine.
Let’s summarize the benefits of Ekkono’s approach:
Ekkono has built a solution that works for all IoT deployments – both industrial and consumer IoT.
For industrial IoT implementations, edge analytics translates into money. Either by being one step ahead of an issue, or one step ahead of the competition:
Ekkono works with various industries like Energy, Telecom, Automotive, Agriculture, Healthcare and Housing Automation. The use cases are commonly within predictive maintenance and production optimization.
For consumer IoT the value proposition is more commonly an elevated user experience through intuitive human-machine interaction. As consumers today, we are more sensitive and susceptible to the overall impression of a product. Adding Ekkono makes your things self-learning about the user’s behavior, and predicts her next step. This facilitates more relevant and applicable tailored recommendations, and better communication with the end-user.
Applicable use cases are home security gateways that can process all the relevant data, learn the normal behavior for the home and its residents, and detect abnormal situations, without sending sensitive data to the cloud; E-bikes that learn the normal use of its owners to provide a smoother and more energy efficient ride; And lawnmower robots that become garden experts from spending thousands of hours hovering your lawn, optimizing cutting schedule, detecting machine issues, and becoming your smart gardener who predicts water and fertilizer needs.
Ekkono is an embedded advanced analytics engine for IoT (Internet of Things) that can execute different machine learning techniques. The unique design makes it resource efficient with an unparalleled small footprint, which enables it to run at the edge, on the actual device. Still, Ekkono does not compromise on functionality:
The design is platform-independent and runs in virtually any environment. It is an all-software solution with no hardware dependencies or requirements. The SDK comes with APIs, input/output of data, bindings to different programming languages, documentation, and tools that help you decorate and prepare data, select machine learning technique, optimize configurations to accommodate purpose, requirements and conditions. This means that you, with programming skills but with no prerequisite of data science knowledge, can get started, implement machine learning on your connected devices, and make them smart.
Ekkono manages the entire workflow from data input, and model training, to model execution, re-training and data output. All these functions are built in a modular fashion that enables one, some, or all of them to run at the edge, in the fog, the cloud, or anywhere in between. The choice depends on hardware configuration, purpose and conditions, like update interval, data divergence and response time. The design allows you to get started and then get better and better as you add functionality and valuable insights about your products, their use and behavior.