*.pcap (zip)

This dataset results from a 47-day Cloud Telescope Internet Background Radiation collection experiment conducted during the months of August and September 2023. A total amount of 260 EC2 instances (sensors) were deployed across all the 26 commercially available AWS regions at the time, 10 sensors per region. A Cloud Telescope sensor does not serve information. All traffic arriving to the sensor is unsolicited, and potentially malicious. Sensors were configured to allow all unsolicited traffic.

Update (07/30/2023): The dataset has been updated to be more realistic with specific characteristics described in [*]. 

We collect encrypted traffic from six widely-used Instant Messaging Applications (IMAs) installed on an Android device for descriptive and statistical analysis, as presented in our papers [*][**]. In particular, we collect traffic from:

1. Microsoft Teams,

2. Discord,

3. Facebook Messenger,

4. Signal,

5. Telegram, and

6. WhatsApp.

This dataset is used for the identification of video in the internet traffic. The dataset was prepared by using Wireshark. It comprises of two types of traffic data, VPN (Virtual Private Network) or encrypted traffic data and Non-VPN or unencrypted traffic. The dataset consist of the data streams (.pcap) of 43 videos. Each video is played 50 times in both VPN and Non-VPN mode. The streams were obtained by setting-up a dummy client on a PC which plays a YouTube video and Wireshark is used to capture the internet traffic.

Smart speakers and voice-based virtual assistants are core components for the success of the IoT paradigm. Unfortunately, they are vulnerable to various privacy threats exploiting machine learning to analyze the generated encrypted traffic. To cope with that, deep adversarial learning approaches can be used to build black-box countermeasures altering the network traffic (e.g., via packet padding) and its statistical information.