Frequently Asked Questions

  • 14 December 2016

Q: In the datasets you refer to V2 while in the rules files Figure 1 - V1 is part of PLC3 and V2 is not connected to any of the PLCs. Please clarify.
A: There is a mismatch between the BATADAL rules and the ‘.inp’ file regarding the labeling of all the valves. V2 is the only actuated valve in C-Town map and is the one connected to PLC3. Readings are available only for this valve. V2 is erroneously named V1 in the rules file.

Q: Do we know which sensor/actuator is being attacked for data labelled as ‘Attack’ in Dataset 4 ? This information was provided on the BATADAL website for Dataset 2.
A: The attack on September 14th and 15th was identified due to pump control inconsistencies in PU10 and PU11. The attacks lasting from October 9th to October 11th, and from October 30th to November 1st, were discovered due to pump PU2 control inconsistencies. The attack on November 27th was directed at PU7, while the abnormal operations from December 6th to December 9th were identified due to pump control inconsistencies in PU6 and PU7. Remember that there might be other undiscovered attacks in Dataset 4.

Q: Are Dataset 1 and 2 still relevant after the release of Datataset 3 and 4?
A: The test dataset that will be released in February will be generated using similar demand patterns to those employed for Dataset 3 and Dataset 4. That said, Dataset 1 and 2 may still be relevant-depending on the approach you are adopting to develop the detection algorithm.

Q: To identify the attacks, can we use all the data provided by EPA Net simulation (demands, pressures, levels, of all elements provided by simulation), or we need to only use data from the sensors reported in datasets (Tanks 1 to 7, Pumps 1 to 11, Valve 2, and pressure nodes J280, J269, etc…)?
A: The sensors deployed in C-Town are only those in the question above (reported in the columns of the datasets). It is possible to retrieve insights on the system by running simulations with EPANET, but beware that the demand patterns provided with the “.inp” file are different from those used to generate the BATADAL dataset. The values of the demands used to generate the available BATADAL datasets are not given.

Q: Is it possible to assume that the potentially attacked measurements coming from the sensors will be given to the algorithm to test for existence of an attack? Or do we have to simulate the actual measurements from the EPA net model?
A: Since the demand patterns used in the simulations are not given, it is not possible to estimate the actual measurements from EPANET and compare them against those we provide to check for anomalies. Participants are asked to use the data regarding “normal operations” (dataset01) and the information on the “.inp” file (network, control logic of the actuators, …) to develop an understanding of the system and produce a detection algorithm.

Q: Is the water distribution system that will be given in February 20 be a different one from C-Town?
A: We will be using C-Town. Apart from the demand patterns, the .inp map will be equal to the one provided in the BATADAL website.

Q: I'm not able to put all my team-mates as Authors. What should I do?
A: You can add your team-mates as Co-Authors. The author is the single corresponding author, you can include as many Co-Authors as you want.

Q: Does the BATAL competition have anything to do with the "Student Competition"?
A: No, and BATADAL contributions cannot be evaluated as part of the student competition.

Q: I cannot find any information about the units of measure represented in the data files. Can you point me to that info?
A: The flow is in LPS. Pressures and water levels are in meters.