FIESTA-IoT Experiments Call
This Call solicits for experiments that design and deploy advanced (experimental) applications, notably applications that will leverage data and resources from multiple administratively and geographically dispersed IoT testbeds.
The scope of the Call is focused on Novel IoT technologies and services. FIESTA-IoT will provide the means for testbed agnostic access to experimental IoT datasets and data-streams, thereby opening new horizons associated with the development of novel/niche IoT technologies and services in areas such as cloud and IoT integration, IoT and Big Data integration, large scale smart cities applications, ambient assisted living environments, management of emergencies and more. Therefore, FIESTA-IoT could allow cutting-edge researchers and innovative enterprises or individuals to develop, validate and test innovative technologies, applications and services, thereby improving their bottom lines. This will be particularly important for SMEs, which do not usually have the resources and equity capital for large scale experimentation.
The main added-value of the FIESTA-IoT platform is that it will provide the opportunity for accessing shared IoT resources, and for using them in the scope of experiments that will combine data from multiple testbeds. The FIESTA-IoT platform offers this interoperability among the datasets from the underlying testbeds employing semantic models and technologies. Experimenters should exploit the semantic and interoperable nature of the datasets and data-streams within their experiments.
A major innovation introduced by FIESTA-IoT relates to the dynamic discovery and use of IoT data from any of the underlying interoperable IoT testbeds.
Benefits for an Experimenter
- The EaaS infrastructure will facilitate experimenters/researchers to conduct large scale experiments that will leverage data, information and services from multiple heterogeneous IoT testbeds, thereby enabling a whole new range of innovative applications and experiments that are nowadays not possible.
- It will enable researchers to share and access IoT-related datasets in a seamless testbed agnostic manner i.e. similar to accessing a large scale distributed database. The objective will also involve linking of diverse IoT datasets, based on the linked sensor data concept. This allows the experimenter to focus on his core task of experimentation, instead of on practical aspects such as learning to work with different tools for each testbed, requesting accounts on each testbed separately, etc.;
- The simplified application process compared to the one from the standard H2020 calls together with a rapid review process by independent external evaluators;
- An extra benefit is the dedicated support from skilled FIESTA-IoT members. This will include their general training on IoT interoperability in general and in FIESTA‑IoT interoperability in particular, targeted consulting services associated with the interoperability of their platforms/testbeds, as well as continuous support in their efforts to use the FIESTA‑IoT results/tools towards improving the level of interoperability of their systems and applications.
Structure of the Call
This call is split in two categories of experiments:
- Scientific excellence targeting experiments validating novel technologies around the IoT concept and its integration with Cloud and Big Data paradigms that clearly advance the current state-of-the-art.
- Innovation targeting experiments validating IoT-based solutions that have a large potential for commercial exploitation in existing or new products or services.
These experiments should be of a short duration (maximum 6 months). Per proposal a budget can be made available up to a maximum of 50 k€ per experiment.
Independent evaluations of the submitted proposals will be performed, in order to select the experiments that will be supported by the project. Different categories of experiments will be evaluated against different criteria (see section 11). It is required that the experiments are performed by a single organization. In the category ‘Innovation by SME’, only proposals from small and medium-size enterprises, including uni-personal companies and individuals, will be accepted.
Data Assembly and Services Portability Experiment
The key feature of this experiment is to build an IoT application that relies on a Smart City Performance Indicator model, based on the information harvested from sensors. Through a set of indicators, the experiment aims at providing the tools to monitor the so-called “health” of a city”. Moreover, this analysis can be split into three different dimensions:
- Detail Level: from general city indicators to specific ones, covering a single aspect of city management, e.g. environmental monitoring.
- Space: from indicators considering the complete city to indicators on the level of places, streets, even houses or rooms.
- Time: ranging from the latest values observed by sensors to the gathering of information that had been observed in the past (i.e. historical data).
These indicators can be used for the visualization, trend analysis and triggering of notifications if a certain situation has occurred. The application will be designed in such a way that different types of sensor information relevant for different application areas can be used. Examples are the monitoring of environmental parameters like pollution, humidity, temperature, light and noise, but could also be: the parking situation in a city/area, water/irrigation levels in a park or agricultural setting, or the activity level in a certain area.
The experiment aims at demonstrating that semantic interoperability across different IoT infrastructures can be achieved, leading to the huge simplification in what respects to the application development.
Dynamic Discovery of IoT resources for testbed agnostic data access
This experiment addresses the assessment and further validation of the capacity to provide an agnostic and seamless access to different assets, provided and supported by potentially heterogeneous testbeds, through the usage of a single Experiment as a Service interface. Namely, the experiment will focus on the dynamic acquisition and processing of information related to the weather/environmental domain (e.g. ambient temperature, air pressure, wind speed, UV, relative humidity, etc.), towards consolidating and visualizing data from multiple locations at the same time. Hence, the experimenter will be able to dynamically specify the locations/areas which data will be collected from, as well as to specify the range of physical phenomena that he/she might be interested in. As can be easily inferred, the specification of these areas/phenomena might lead to query data from one or more testbeds; however, the process of gathering all this information is completely seamless to external users.
Large Scale Crowd-sensing Experiments
This experiment will setup and execute a range of crowd-sensing trials through the harvesting of data coming from citizens’ handheld devices, following the “Living Lab” paradigm. This experiment will focus on the noise within the context of a large-scale environment. With this, experimenters will be able to identify and even predict noise variations (both spatially and temporally).
The major goal of this experiment is to explore the ability of FIESTA-IoT platform to manage data coming from different sources either mobile or static. This experiment will specifically utilize data available in FIESTA-IoT platform that is made available to FIESTA-IoT via participatory sensing approach and static sensors available in the region of interest.
Predictive Traffic Condition Analytics
The following experiment aims to check whether a traffic monitoring prediction system can be scalable enough through the utilization of commercially off-the-shelf tools. Its main goal is to have a Hadoop Distributed File System (HDFS)-based setup, which gathers real-time traffic sensor data. Whilst the data is being harvested, a set of machine-learning tasks are executed, detecting potential incremental changes (e.g. through Weka or PredictionIO, well-known data mining and machine learning frameworks). Based on these deltas, a set of machine-learning algorithms will predict the near-term future, e.g. future traffic congestions, emergency routes, etc. A separate process will periodically examine the entire dataset to devise future traffic patterns (e.g. through R). The overall objective of the experiment is therefore to discover if it is possible to have a functional framework, built upon commercially off-the-shelf tools that can perform predictive analysis in near real-time on a large dataset, populated via sensors in an “Internet of Things” scenario, like the city of Santander.
Environmental Noise Monitoring using Acoustic Data
Europe is considered to have one of the most restrictive and extensive environmental laws in the world. The environmental policies in Europe seek to be as much environmental-friendly as possible, increasing their citizens’ quality of life as much as possible while leading the fight against the upcoming environmental challenges. This experiment, which runs over the IoT facilities providing noise monitoring information, aims to create a real-time noise monitoring map using acoustic data which can even substitute expensive off-the-shelf offline noise map techniques. The acoustic data as it is, is not sufficient to give a precise noise map, because there is coarse spatial sampling of the urban environment by acoustic/noise sensors, which prevents their combination and visualization in a suitable form. It requires post-processing of the sensor data to improve and get appropriate data. However, the measurements are highly dependent on several factors:
- Location and proximity to noise sources.
- Malfunctioning sensors.
- Sensor calibration to a common reference.
- Sensor dynamic range (which is slightly variable among them while gathering data but all of them are able to measure between 50 and 100 dBA).
Using data gathered from acoustic sensors that fulfil the aforementioned factors, a regression analysis can be performed, thus calculate the second order polynomial function to all the data measurements. The representation of the noise map in real time can be done using a straightforward web interface.