Back to search

IKTPLUSS-IKT og digital innovasjon

Autonomous Resource-constrained Things (ART)

Alternative title: Autonomous Resource-constrained Things (ART)

Awarded: NOK 0.59 mill.

In the near future, many applications within the Internet of Things will depend on data delivered by myriads of sensors. Examples are monitoring emissions within a city on a very detailed level, or monitor the usage of the facilities within a building or production plant. The sensor devices needed for such application must be tiny and wireless, so that they can be installed easily everywhere, without any cables and at low costs. This constraints how much energy they have at their disposal. Some of them are battery-driven, others use energy-harvesting techniques such as small solar panels. For that reason, they must be designed and programmed very carefully to use energy very efficiently. We found, however, that each sensor will find itself in very different environments, for instance a sensor that only gets little light, or another sensor that is installed at a very interesting place that requires it to measure much more often. Because there are so many sensors, it is impossible for engineers to optimize each of them individually. Instead, the sensors have to learn about their environment on their own and optimize over time. For this, machine learning algorithms can be used, that for instance decide about the sensor's strategy based on the upcoming weather forecast, or the current traffic situation within a city. In this project, we work on approaches that allow sensor devices to benefit from machine-learning despite their constraints. They should be able to adapt their behavior to the given situation, and thereby optimize the overall application behavior. This enables cost-effective installation and operation of sensing applications in a wide variety of use cases.

ART aims at providing methodology for autonomous management and cost-effective ubiquitous IoT by adapting and developing innovative big-data and machine-learning techniques. The goal of the concluding part of the project is to communicate results of phase 1, and to verify how they can be used for new project proposals and other application areas.

Funding scheme:

IKTPLUSS-IKT og digital innovasjon