Find research datasets worth reusing

opencc-by-4.0Jun 2016View details →

BRAIN Journal-Participative Teaching with Mobile Devices and Social Networks for K-12 Children-Figure 11. Children applying weaving techniques under the control of the staff from NUA Bucharest

The information from a) and b) has been used by other colleagues in the local school and in the future will be used by other schools in the country or abroad. This represented the second educational level of the project, i.e. the analysis of the collected data. Information from c) was further analyzed by university teachers, filtered and added as an enhancement to the content of the AR platform. An example of user-created content are the movies made with smartphone cameras, recording the children2 while performing traditional crafts (Figures 11, 12).

BRAIN Journal-Image Finder Mobile Application Based on Neural Networks-Figure 11.Weka Results for Images

The dataset of the real images is also treated the same way as the dataset of the sketches. Total tree images used = 86 Total car images used = 54 Learning rate = 0.3 Momentum = 0.2 Number of epochs = 500 70% of data is used to train the neural network and the remaining 30% is used for testing the trained neural network. Figure 10 shows the neural network for images. The results are shown in Figure 11 and are as follows: Total Correct Recognition = 83.7838% Total Incorrect Recognition = 16.2162% Error Per epoch = 0.0235851

BRAIN Journal-Image Finder Mobile Application Based on Neural Networks-Figure 8. Training Neural Network for Sketches

It is described in the previous section that a mobile application is used to collect data about the sketches and the data is converted to 20x20=400 integer numbers to give it as input to Weka. Our experimentation includes only two objects for recognition i.e. trees and cars. Total tree sketches used = 175 Total car sketches used = 72 Learning rate = 0.3 Momentum = 0.2 Number of epochs = 500 70% of data is used to train the neural network and the remaining 30% is used for testing the trained neural network. Figure 8 shows the neural network for sketches. The results are shown in Figure 9 and are as follows: Total Correct Recognition = 100%

BRAIN Journal-Image Finder Mobile Application Based on Neural Networks-Figure 5. Cropped Sketch

Figure 5 shows the interface of the mobile application that is created to collect the data of the sketches. It contains a drawing canvas; where drawings are made and it also contains few text boxes which are filled with the information of the drawing. For example, if the drawing is a tree, ‘1’ is written in the tree text box. Furthermore, the undo and redo buttons help modify the drawing and the clear button clears everything drawn onto the drawing canvas.  The drawing of a tree is shown on the drawing canvas in Figure 4. It is clear from the figure that there is some empty area on top, right, left and bottom of the drawing sketch, which can cause problems while using this image for training the neural network. Therefore, there is implemented a crop function which crops the image very minutely and removes the empty space as shown in Figure 5.

BRAIN Journal-Image Finder Mobile Application Based on Neural Networks-Figure 4. Mobile Application to collect Sketch data

The drawing of a tree is shown on the drawing canvas in Figure 4. It is clear from the figure that there is some empty area on top, right, left and bottom of the drawing sketch, which can cause problems while using this image for training the neural network.

BRAIN Journal-Image Finder Mobile Application Based on Neural Networks-Figure 3. The used methodology

The training process we use in both trainings is based on neural networks, using the backpropgation algorithm. By using this algorithm, we are going to finally obtain the weights that will be used in our model to make the system recognize the input given by the user.  Once the training for the hand drawn sketches is over, we are going to get weights that will be used to recognize any inputted hand drawn sketches. For example, if the user draws a tree, the system will be able to recognize what has been drawn as a tree, using the obtained weights from the neural networks. On the other hand, once the training for the real images is over, it means that if we provide our system with a real image, for example a tree, the system will be able to recognize it using the obtained weights from the training phase.

BRAIN Journal-Image Finder Mobile Application Based on Neural Networks-Figure 6. Reducing size of Sketch

As previously described, there are a total of 1600 integers, which can be the input to the dataset for neural networks. But it is a huge number, so in order to minimize the size of inputs to the neural network the 40x40 matrix is reduced to 20x20 by skipping odd rows and columns of the original matrix. Figure 6 shows a matrix containing green and red rows and columns. If this was the 40x40 matrix, then the red part of this matrix would be skipped to convert it into a 20x20 sized matrix. Now there are only 20x20=400 values, which is a reasonable input size for the neural network.  The other part of the developed approach is to collect the data about the same two objects of the real pictures taken by the camera. These images are converted into black and white pictures and then treated the same way as the sketches, i.e. black and white pictures are also converted into integers based on the color of each pixel.

BRAIN Journal-Image Finder Mobile Application Based on Neural Networks-Figure 1. Methodology used in the research by (Egmont-Petersen et al., 2002)

Figure 1 shows the methodology used by Egmont-Petersen et al. (2002) to come up with an answer to their research question. Their study says that image recognition using neural networks goes through the stages shown in Figure 1. In our research we shall use this general proposed approach.

BRAIN Journal-Image Finder Mobile Application Based on Neural Networks-Figure 10. Training Neural Network for Images

BRAIN Journal-Image Finder Mobile Application Based on Neural Networks-Figure 2. The used approach

In this section we present in details the methodology used for this research. Figure 2 depicts the used methodology. The adopted approach is based on first training the system to be able to recognize certain sketches by providing it with various hand-drawn examples such as trees, cars and mountains. Secondly, the system has to be trained again to recognize real images by providing it with real images such as trees, cars and mountains.

BRAIN Journal-Image Finder Mobile Application Based on Neural Networks-Figure 7. Cropping image

This function takes five parameters, a WriteableBitmap type of object, the location of the starting point on x coordinate, the location of the starting point on y coordinate, width and height. It crops the image according to these parameters as shown in Figure 7.

BRAIN Journal-Image Finder Mobile Application Based on Neural Networks-Figure 9. Weka Results for Sketches

the sketches and the data is converted to 20x20=400 integer numbers to give it as input to Weka. Our experimentation includes only two objects for recognition i.e. trees and cars. Total tree sketches used = 175 Total car sketches used = 72 Learning rate = 0.3 Momentum = 0.2 Number of epochs = 500 70% of data is used to train the neural network and the remaining 30% is used for testing the trained neural network. Figure 8 shows the neural network for sketches. The results are shown in Figure 9 and are as follows: Total Correct Recognition = 100%

opencc-by-4.0Jul 2019View details →

Ozone and Carbon Monoxide Dataset Collected by the OpenSense Zurich Mobile Sensor Network

Ozone and Carbon Monoxide Dataset Collected by the OpenSense Zurich Mobile Sensor Network This dataset contains ozone (O3) and carbon monoxide (CO) concentration measurements collected by the OpenSense (<a href="http://www.opensense.ethz.ch">http://www.opensense.ethz.ch</a>) mobile senor network over the course of 4.5 years (2012/02-2016/09). The sensors are mounted on top of 10 streetcars in the city of Zurich, Switzerland. In particular, the dataset contains:  <ol> <li>Ozone (O3) data: 2012/02 - 2016/09 (19.9 Mio samples)</li> <li>Carbonmonoxide (CO) data: 2014/03 - 2016/09 (49.7Mio samples)</li> </ol> Hardware: -------------- <ol> <li>Ozone sensor: SGX (former e2V) MiCS-OZ-47 Ozone Sensing Head with Smart Transmitter PCB</li> <li>Carbon monoxide sensor: Alphasense CO-B4</li> <li>GPS receiver: u-blox EVK-6p</li> </ol> Data files format:  ------------------------- co_data_*:  <ol> <li>Time of day: yyyy.mm.dd HH:MM</li> <li>Latitude WGS84</li> <li>Longitude WGS84</li> <li>HDOP: horizontal dilution of precision, uncertainty of the GPS position</li> <li>Tram ID</li> <li>WE_CHANNEL_SENSOR_1_MV: The voltage [in mV] at the working electrode of the electrochemical sensor (see Alphasense CO-B4 datasheet for more details)</li> </ol> o3_data_*:  <ol> <li>Time of day: yyyy.mm.dd HH:MM</li> <li>Latitude WGS84</li> <li>Longitude WGS84</li> <li>HDOP: horizontal dilution of precision, uncertainty of the GPS position</li> <li>Tram ID</li> <li>Ozone [ppb]: On-device calibrated (according to manufacturer) ozone measurement [in parts-per-billion] </li> <li>Temperature [in °C]</li> <li>Relative Humidity [in %]</li> </ol> Data quality: ------------------ The data has NOT been post-processed! In order to achieve high data quality, the data needs to be cleaned (e.g. outlier filtering) and, most importantly, the sensors need to be individually calibrated. Reference data can be obtained from <a href="http://www.ostluft.ch">www.ostluft.ch</a>, the official air quality monitoring network in eastern Switzerland, which operates multiple monitoring stations in the city of Zurich. Plot Coverage Map (MATLAB): -------------------------------------------- The provided MATLAB script plot_data_coverage.m plots the locations of the collected samples onto the map of Zurich (map_zurich.png).  References: ----------------- The dataset (and related aspects) has partly been used and is described in more detail in the following publications: <ol> <li>Balz Maag et al. SCAN: Multi-Hop Calibration for Mobile Sensor Arrays. In Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, Vol.1, No.2 (IMWUT), 2017.</li> <li>Olga Saukh et al. Reducing Multi-Hop Calibration Errors in Mobile Sensor Networks. In IEEE/ACM International Conference on Information Processing in Sensor Networks (IPSN), 2015. Best Paper Award!</li> <li>Olga Saukh et al. Route Selection for Mobile Sensor Nodes on Public Transport Networks. In Journal of Ambient Intelligence and Humanized Computing, 5(3), Springer, 2014.</li> <li>Olga Saukh et al. On Rendezvous in Mobile Sensing Networks. In Proceedings of the 5th Workshop on Real-World Wireless Sensor Networks (RealWSN), 2013.</li> <li>Jason Jingshi Li  et al. Sensing the Air we Breathe – The OpenSense Zurich Dataset. In Proceedings of the 26th International Conference on Artificial Intelligence (AAAI), 2012.</li> <li>Olga Saukh et al. Route Selection for Mobile Sensors with Checkpointing Constraints. In Proceedings of the 8th International Workshop on Sensor Networks and Systems for Pervasive Computing (PerSeNS, in conjunction with IEEE PerCom), March 2012.</li> <li>David Hasenfratz et al. On-the-fly Calibration of Low-Cost Gas Sensors. In Proceedings of the 9th European Conference on Wireless Sensor Networks (EWSN), 2012. </li> </ol> For further information, visit:  <a href="http://www.opensense.ethz.ch">http://www.opensense.ethz.ch</a>

opencc-by-4.0Nov 2021View details →

Synthetic dataset for mobile wireless networks with SUMO -- aggregated traces

Aggregated dataset from a published wireless dataset generator base in SUMO mobility model.    This work was supported by national funds through Fundação para a Ciência e a Tecnologia (FCT) with reference UIDB/50021/2020 and SFRH/BD/132053/2017.

opencc-by-4.0Sep 2021View details →

Dataset for Agile 5G Network Measurements: Operator Benefits of Employing Aerial Mobility

Dataset of paper "Agile 5G Network Measurements: Operator Benefits of Employing Aerial Mobility". The measurement data contains two different scenarios: Scenario 1 and Scenario 2, according to the paper. These two datasets include two different kinds of files: those related to measurements taken with a phone and those taken with a drone (UAV).

openother-openFeb 2021View details →

PatchFlow v1.0 mobility networks

Initial release of the synthetic mobility networks calculated using radiation model based on GPW-GADM admin 1 and admin 2 level regions

opencc-by-4.0Oct 2022View details →

DAEMON: Network intelligence for aDAptive and sElf-Learning MObile Networks

The DAEMON H2020 european project develops and implements innovative and pragmatic approaches to Network Intelligence (NI) design that enable high performance, sustainable and extremely reliable zero-touch network system. DAEMON designs an end-to-end NI-native architecture for Beyond 5G (B5G) that fully coordinates NI-assisted functionalities. Main website: <a href="https://www.youtube.com/redirect?event=video_description&redir_token=QUFFLUhqblFJekN4dzdUcHYzQXlMd19rS0xjSDNleWVlZ3xBQ3Jtc0tsSmpkdWVUaXBSVDRHb2w5WDhGZnN3S05DM0FSRk9VS3RDc1hTT0l6MVhtd1pvWmZ2bEtRRUFxb0xkRmRQa1lVWVF0a1B2M1RwdXlmS2lIWEF0LWt4Q1ZIQndSNHNJcVJaSVNQODgtRThEaHh4TDBuaw&q=https%3A%2F%2Fh2020daemon.eu%2F&v=VNMVe8S-Ees">https://h2020daemon.eu/</a>

opencc-by-4.0Apr 2024View details →

Distributed predictive QoS in presence of network- and mobility-related drifts

The datasets represent a dynamic environment, where several client-vehicles are moving in an urban area. Each client runs a streaming cloud service constantly receiving data packets. Network simulation is performed using Simu5G, a library that emulates a 5G cellular environment in OMNeT++. The simulator's radio parameters are set according to the Macro-cell model proposed by International Telecommunication Union. The map comprises of an urban 600x600 square meters area located in a suburb of a European capital. Inside this area four 5G base-stations (gNodeBs) have been installed by the national network operator, enabling four 5G cells. This area, divided into several blocks by the actual road network is integrated in our simulation by an OpenStreetMap (OSM) instance. The total number of included vehicles is set to 25, according to vehicle density statistics in the corresponding country. The road network's traffic is simulated by SUMO that creates a digitized version of the (real-world) OSM map and produces the route files for the vehicles. Route files are loaded in the Simu5G simulator, where a network-vehicular mobility co-simulation takes place. For each vehicle's route we assume SUMO's default parameters for urban environment: exponential speed model (with maximum speed restriction as defined by the OSM traffic rules) and the probability matrix at intersections for {lane keeping, turn left and right} as {0.5, 0.25 and 0.25}, respectively. The following information is collected for each vehicle using OMNeT++'s monitoring service: timestamp, channel quality indicator, packet delay, measured signal to noise ratio (SNR), client position (x,y,z), client velocity (x,y,z), received SNR, radio link control throughput, serving cell, client throughput. These features are sampled at 1 Hz and comprise the values of our synthetic time-series QoS dataset.  We have created two drift datasets that correspond to complementary cases of major long-term changes in the considered environment: 1) a network infrastructure-driven scenario (Sc1) and 2) a human behavior-driven scenario (Sc2). In Sc1 we assume that two out of four gNodeBs are switched off under a cost-reduction on/off policy or an infrastructure-share strategy (adopted by MNOs) that would imply such changes. For Sc2 we modify the users' mobility pattern; we assume that a "hotspot" e.g., a metro station is created in the lower-right edge of the map resulting in a traffic increase to that area. This is achieved by increasing the probabilities of the routes leading to the "hotspot" in SUMO's route planning. All generated datasets have a total duration of 20 hrs (simulation time) and the respective drift event is introduced at t=10 hrs.

opencc-by-4.0Jul 2024View details →

Validation Videos - Robotic System for Reproducible Mobile Networking Experimentation in Anechoic Chambers (Master Thesis)

Note on Robot's Referential: The robot's referential can be inferred in the recording via the "Safety Position." The safety position is the same for both the Digital Model (Gazebo) and the Real Robot (Joint Position = [0.0, -1.57, 1.57, 0.0, 0.0, 0.0]). In the safety position, the robot is approximately aligned with the X-axis, with its end-effector on the positive side of the axis. The end-effector faces perpendicular to the Y-axis. The positive Z-axis points upwards towards the ceiling.