Internet of Things: A pilot deployment in the city of Melbourne, Australia

As part of the 'Internet of Things (IoT) for Creating Smart Cities' project, the ISSNIP group of the University of Melbourne, ARUPand the City of Melbourne council (CoM) have done a pilot deployment of IoT networks, in the Melbourne city, for monitoring environmental parameters. One of the aims of the research is to develop new systems and algorithms that can help City administrators remotely monitor, understand and interpret real time information on urban environments. The environmental sensors, measuring light levels, humidity and temperature, have been deployed at Fitzroy Gardens and Library at the Dock . The data collected will assist the Urban Landscapes branch of CoM to better understand and communicate the impact of canopy cover for urban cooling.

Real time Data:
There are five sensors deployed at the Fitzroy Garden and four sensors deployed at the Docklands library. The data is collected at 10 minutes interval and the live data is available from the open data platform of CoM. The link is provided below:

Please cite the following two papers when using the data:

[1] Alistair Shilton, Sutharshan Rajasegarar, Christopher Leckie, Marimuthu Palaniswami " DP1SVM: A Dynamic Planar One-Class Support Vector Machine for Internet of Things Environment", In proceedings of the International Conference on Recent advances in Internet of Things (RIoT 2015), April, 2015.

[2] Jayavardhana Gubbi, Rajkumar Buyya, Slaven Marusic, and Marimuthu Palaniswami, Internet of Things (IoT): A Vision, Architectural Elements, and Future Directions, Future Generation Computer Systems, Volume 29, No. 7, Pages: 1645-1660, ISSN: 0167-739X, Elsevier Science, Amsterdam, The Netherlands, 2013.

Some news about the deployment:

Location of the sensors:

Network diagram of the sensors:

Docklands Library Deployment:


Fitzroy Garden Deployment:

A graph of temperature measurements collected by the Docklands Library sensors between 21st Dec 2014 and 1st Feb 2015:

Visualisation Tools for Monitoring Real-Time Data  

The below two links provide video demonstrations of the visualisations created to view and analyse the real time data collected from the above deployment.

Visualization-1:  In this visualisation video, information about the tree species, canopy cover percentage and the numeric value of sensor readings (humidity, temperature and light) are displayed. This helps to relate different canopy cover and tree species with the climate parameters. Canopy coverage is also shown using a shadow of the tree that changes its width/size depending on the percentage of coverage. A sliding bar is included, for the user, to vary the data/time manually and visualize the microclimate at that instance. Click on the image to start the video.

Visualization-2:  This visualisation video shows the estimated parameter values (temperature, humidity or light) at a selected location pointed/selected by the cursor over a spatial region, where the sensors are deployed. Click on the image to start the video

IoT Experiment on "Adaptive Cluster Tendency Visualisation and Anomaly Detection for IoT Environment"

The below three links show videos of the adaptive clustering tendency visualisation in action. Three datasets are used: namely, (1) Melbourne IoT Data (as detailed above in this web page), (2) Heron Island data  and (3) IRIS benchmark data (UCI Repository)

1) iVat image movie for Melbourne IoT Data

We use the light levels, humidity, and temperature data obtained from four sensors at the Docklands library for the sliding window based clustering experiment. We re-sampled the dataset so that we have 1 measurement of temperature, humidity, and light from all four sensors every 30 minutes. These measurements are concatenated to obtain a 12 dimensional feature vector (4 sensors × 3 measurements per sensor (temperature, humidity, and light)). In total we have measurements for roughly 72 days. We perform a window based clustering experiment on this dataset with a window size of 2 days (2 days × 24 hours per day ×2 samples per hour = 96). So incVAT/inciVAT adds sample points to the MST for the first 96 samples and then applies one decVAT/deciVAT operation to remove the oldest datapoint and one incVAT/inciVAT operation to insert the new datapoint into the MST. At the last step, we use only decVAT/deciVAT to bring the number of datapoints up to 2.

The video (see below) shows the plot of temperature, humidity, and light for each of the sensors, the MST cut magnitude (dn) and inciVAT/deciVAT image based on Euclidean distances in R12, showing a visual estimate of the number of clusters at each instant of data collection. We use the parameters α = 6 and β = 0.1 to perform the clustering and anomaly detection task. The anomalies are shown by black dots and clusters are shown by different colours in the plot of temperature, humidity, and light for all the sensors.

Click HERE for the video

2) iVat image movie for Heron Island Data

This experiment is a real dataset collected from the Heron Island weather station deployed on the Great Barrier Reef, Australia. The Heron Island data has three variables: (air) humidity, (air) pressure, and (air) temperature. The data were collected every 10 min from 9:00 am to 3:00 pm each day for the 30 days beginning 21/2/2009 and ending 22/3/2009.For this experiment, we have set the parameters α = 15 and β = 0.04. The normal datapoints are shown by blue dots, whereas two (separate) anomalies are shown by red and green dots respectively in the 3-D scatterplot of the Heron island data. The anomalies are represented by red and green square block along the diagonal of the inciVAT image.

Click HERE for the video

3) iVat image movie for iris Data

In this experiment we deliberately inserted five anomalies at random locations into the IRIS data to see whether incVAT spots them as they occur. Specifically, we inserted (10,10,10,10), (-10,10,10,10), (-9.6,10.2,-10.7,8.9), (10,10.2,9.8,10), and (-10,10.2,9.8,10) at locations 21, 35, 70, 94, and 130 respectively. For this demonstration, we have set the parameters α = 10 and β = 0.02. The video shows the inciVAT image of IRIS data with 5 anomalies. The anomalies are shown by red, green and magenta pixles.

Click HERE for the video

top of page