The embedded systems group developed their own ecosystem for the Internet-of-Things (IoT): The Elastic IoT Plattform (EIP). The EIP leverages the concept of utmost decoupling of IoT devices and their controlling software. This decoupling allows for great flexibility during development, maintenance and even deploying of IoT systems. This project focuses on the the software abstraction of devices within the EIP, so called "digital twins". The aim of the project is developing multiple digital twins and their interactions to delve deeper into the EIP and its capabilities.

The first of half of digital twins will simulate existing IoT systems. Due to the current limitations using our own systems is sadly not possible. In its stead we will use web APIs of existing, deployed systems. e.g. weather information, power system information etc. The digital twins of these APIs will be treated as normal devices in the EIP.

The second half of twins will focus on processing the data. Techniques from "unsupervised learning", especially "deep neural networks" will be used. One possibility would be data compression, but the project is not limited to it.

Requirements

The EIP is implemented in Java. Development of digital twins will use Java, or at least a JVM based language. Appropriate experience is a must. The EIP is in its early development stages and is currently developed in a bleeding edge fashion, though working. Bugs and minor inconveniences are to be expected. General experience in programming, especially a systematic approach to debugging are beneficial. Prior experience in unsupervised learning and deep neural networks is beneficial but not necessary. 

This project is limited to BAI/MAI and ISE M-CE.

Digital Course:

This course will only be offered digitally, through videos and live streams. To participate in this couse, please enroll into the corresponding Moodle-Course. To enroll in the Moodle-Course this semester you do not need a password. In the course you will find all necessary informations to be able to participate in the first session. The kickoff-meeting will start on Monday, 20.04. at 12:00h. Participation in the kickoff-meeting is mandatory to join the project.

Embedded are only really useful if you can communicate with them. Since many of them are battery powered and not physically connected, this needs to be wireless. Our embedded platforms in the ES lab (the so-called Elastic Node and motherboard), use a low-powered IoT standard called 802.15.4. We want to expand this for newer and faster ARM-based devices, adding the more universal solutions of WiFi and Bluetooth. This will allow the devices to communicate directly with laptops and phones, and take advantage of the larger messages and faster transfer rates than 802.15.4.The focus of this project is to design a board that extends the original motherboard embedded design to include either a combinational bluetooth/wifi module, or dedicated ones. Therefore, some experience with electronic circuit design is a requirement, while experience with embedded systems is a strong suggestion. PCB design using software such as Eagle would be beneficial, but is not required.

This project is limited to BMAI/MAI and ISE M-CE.

Digital Course:

This course will only be offered digitally, through videos and live streams. To participate in this couse, please enroll into the corresponding Moodle-Course. To enroll in the Moodle-Course this semester you do not need a password. In the course you will find all necessary informations to be able to participate in the first session. The kickoff-meeting will start on Wednesday, 22.04. at 14:00h. Participation in the kickoff-meeting is mandatory to join the project.