 Hello everybody, this is Manuel Cantone with ST Microelectronics, IoT Marketing Manager in charge of cloud connectivity, and we are here at the IoT World 2019 to show a predictive maintenance platform demo, an end-to-end example on how to post sensor data all the way to the cloud. In this example, the cloud of choice is AWS, as you see, and we're going to start with a rig, a mock-up of an industrial equipment with two identical motors retrofitted with our sensors that post the sensor data to the cloud. So we have here our two motors, which are retrofitted with two evaluation kits from ST Microelectronics available on our website. They are little sensor sticks that are composed by an STM32 controller, part of our F4 family, so a Cortex, an ARM Cortex-M4, they have an accelerometer on board, a gyroscope, a microphone and a temperature and a humidity sensor. In this case, we are monitoring environmental data and vibration data. These data are pre-processed on the STM32, and an FFT is performed to minimize the amount of data that needs to be sent up to the cloud, and they're sent through IO link, an industrial connectivity link, to this master board that translates into a serial so that we can send the data through USB to a gateway. This is the STM32 MP157C DK2, our discovery kit for the STM32 MP1, and it's single board computing as you see, it's very developer friendly and excellent as a starting point to get your data in development. In this case, it's running AWS Greengrass, and it's taking the sensor data sent through the serial communication and creating the objects, the things that are going to post data messages via MQTT to the AWS IoT Core. See in this slide here the main architecture, so the point of ingestion is the AWS IoT Core and we instantiate from AWS to do storage, like Amazon S3, to do function with the Lambda function and others to manage the security of the dashboard. Once the data are on the cloud, we can visualize the data on this dashboard, and what we can observe here are the vibration data, again, frequency, as the data are processed on the sensor node, and environmental. So I see these two nodes that are monitoring the identical motor, and I can observe a different amplitude on the motor on the left. This is because we purposely offset the rotor of the motor on the left with a nut and bolt, so it will experience a higher amplitude on the lower frequency, simulating a signature for an early failure of an equipment. We can do some basic condition monitoring, putting thresholds on the various objects and creating alerts or even alarms and notification sensors from the dashboard all the way to the end user. These are now running on the cloud. Thanks to AWS Greengrass, they can run also on the gateway, minimizing the latency and creating a faster action upon certain happening. This dashboard allows you also to augment the data from the sensor with geographic position data, so we can see our node zero at IoT World 2019, one of them as a warning, and that's what we just observed in the previous tab. Going back to the dashboard, you can observe a third node. It's this discovery kit that you see over here. This is the discovery kit IoT node. It's our to-go board as a starter kit for cloud development. It runs Amazon Friartos, and it connects directly to the AWS IoT Core via Wi-Fi sending MQTT messages. This is sending environmental data, and again, it's to show that this can be done wired and wirelessly with a STI product together with AWS services. This is a platform to ingest data, so we're showing how to go to sensor to cloud. Once the data is in the cloud, you can imagine this rig multiplied by hundreds of equipment and maybe tens of different fabs also in the same equipment. The idea is that on this data, you can do some analysis and create some insight to identify early signature for failures. We provide tools to do that, like for example, STM32QBI, but we don't provide the actual prediction of failure. So we will release this platform as an application node soon on ST.com. And you will be able already to source all the components from all the development kits that we are showing here, so the developers can actually duplicate this, start ingesting data, observe the dashboard and start building actual applications and creating insight and ultimately value out of the data. Another important point is that from the cloud, we can update the behavior of the sensor node. And as I was executing lambda function on the gateway, I can also push alerts and behavior all the way to the sensor node. I do this to over the update on the Discovery Kit node running Amazon FreeRTOS, I can update the firmware also on the sensor node on this rig. Thanks for watching. This was Manuel Cantone, and for any information, go to www.st.com. Thanks for watching.