Table 4.

List of various cyber-attack datasets.

Reference	Year	Dataset Used	Dataset Size	Format	Details about the Dataset/Brief Description
[104]	2016	CAIDA DDoS 2007 and MIT DARPA dataset	5.3 GB	pcap (tcpdump) format	-An hour of anonymized traffic records from a DDoS attack on 4 August 2007 is included in this dataset. -This form of denial-of-service attack tries to prevent users from accessing the server by using all of the computational resources and bandwidth on the network.
[105]	2015	Botnet [Zeus (Snort), Zeus (NETRESEC), Zeus-2 (NIMS), Conficker (CAIDA) and ISOT-Uvic]	14 GB packets	packet	-It is a network-based dataset. It basically works on diverse networks and intercepts emulated traffic (Normal and attack traffic). -The data set is well labelled but not balanced.
[106]	2009	NSL-KDD	4 GB of compressed (approx.)/150k points	tcpdump data	-The train set does not contain any redundant records nor any duplicate records. -A limited number of datasets are taken into consideration for training and testing.
[107]	2011	ISOT	11 GB packets	packet	-The ISOT dataset was compiled from two different datasets comprising malicious traffic from the Honeynet project’s French chapter, which involved the Storm and Waledac botnets, respectively.
[54]	2016	UNSW-NB-15	100 GB	CSV files	-The Australian Centre for Cyber Security’s (ACCS) Cyber Range Lab used an IXIA PerfectStorm programme to construct a combination of realistic modern regular activities and synthetic contemporary attack behaviours from network data.
[108]	2017	Unified Host and Network	150 GB flows (compressed)	bi. flows, logs	-A subset of network and computer (host) events makes up the Unified Host and Network Dataset, events were collected over a 90-day period from the Los Alamos National Laboratory enterprise network.
[109]	2011	Yahoo Password Frequency Corpus	130.64 kB (compressed)	txt files	-The dataset contains sanitised password frequency lists from Yahoo, which were obtained in May 2011.
[110]	2014	500K HTTP Headers	75 MB	CSV files	-Crawled the top 500K sites (as ranked by Alexa).
[111]	2014	The Drebin Dataset	6 MB (approx.)	txt log, CSV and XML files	-The goal of the dataset is to promote Android malware research and allow for comparisons of different detection methods. -There are 5560 applications in the dataset, representing 179 separate malware families. Between August 2010 and October 2012, the samples were collected.
[112]	2008	Common Crawl	320 TiB	WARC and ARC format	-Since 2008, the Common Crawl corpus has accumulated petabytes of data. -Raw web page data, extracted metadata, and text extractions are all included, composed of over 50 billion web pages.