OctoAI - Next Generation of High Performance Edge AI für Smart Buildings

At present, the building stock in the EU remains energy intensive and mostly inefficient; it is responsible for 40% of final energy consumption and 36% of CO2 emissions. In order to increase the share of renewable energy and reduce energy consumption, future systems must have a high degree of flexibility and efficiency. On the one hand, this requires the systematic embedding of cyber technologies in order to monitor the physical systems and enable communication between different subsystems. On the other hand, innovative energy services such as demand-side management or model-predictive control are required to reduce the energy consumption of buildings and to transform buildings into active, intelligent players in higher-level energy systems. Studies have shown that artificial intelligence (AI - Artificial Intelligence) is the backbone and enabler of many energy services. Key Message 1: Applications of artificial intelligence are the backbone and enabler of many energy services. Innovative Energy Services are built on a bi-directional, real-time interaction with real buildings. Innovative solutions are required for the generation, provision and evaluation of these large amounts of data. Internet of Things (IoT) technologies are the backbone and an enabler of these intelligent systems. Computing paradigms Current IoT implementation depends almost entirely on cloud infrastructure and cloud-based services. Cloud computing offers numerous advantages such as cost efficiency, high availability, inexpensive software and enhanced security [6]. However, cloud-based services also have serious disadvantages: reliability, trustworthiness, or security and data protection. These disadvantages result, among other things, from the fact that the data provider (end user) and the data consumer (cloud provider) often have conflicting interests. Key Message 2: In the area of ​​energy services for intelligent buildings, cloud computing has serious weaknesses in the areas of reliability, trustworthiness, and data protection/security. Edge computing is an alternative IoT implementation and refers to computing taking place at the edge of networks; the "edge" is where end devices access the rest of the network. Edge computing increases availability, accessibility and reliability and improves latency for many services compared to cloud computing applications. Users usually own the end devices and have physical access on site to control them. This also increases user trust as data migration becomes optional for many use cases, which in turn leads to a lower risk of data breaches. Key Message 3: In principle, edge computing can overcome the main problems and limitations of cloud computing. Edge devices have power consumption limitations and therefore have limited computing resources. Cloud-based AI applications usually consume a lot of energy and cannot (or only to a very limited extent) be used on resource-constrained devices. A central challenge for edge applications in the field of intelligent buildings is to "bring AI to the edge". Key Message 4: In order to use the full potential of edge computing for intelligent buildings, "AI must be brought to the edge" of networks.