In the initial version of PodNet, content publication was done anonymously, making it an ideal platform for spreading spam and illegal content. PodNetSec is the secure version of PodNet. It implements mechanisms – leveraging real world social networks and communities – to prevent the dissemination of spam and illegal content. A two-level rating system as well as social trust metrics based on friend ties and local communities, are introduced and serve as input of a reactive as well as a proactive spam control mechanism.

PodNetSec introduces 3 types of channels:

• Closed channels allow private and encrypted dissemination of content in a limited group e.g., for friends to share photos.

• Restricted channels only allow authorized users to publish content but everybody to consume it, thus basically supporting official podcasters (e.g., BBC).

• Open channels allow every user to consume as well as create new content. This is useful for applications such as discussion forums, or video sharing (e.g., YouTube), but may serve as an easy platform to spread spam and objectionable content. Open channels hence rely on a reputation system, spam control mechanisms, and trust metrics (all detailed below).

A two level rating system is introduced to allow for content to be rated and author’s to build up a reputation (based on their publications). It allows to rate, first, the legitimacy and, second, the quality of the content. Ratings are then shared among users for a more accurate and faster creation of the reputation. The received ratings are weighted by the raters trust value in order to cope with liars. Whenever content is legitimate, a user may assess their satisfaction with it, in order to tune future download selection to their taste. Otherwise, a user can declare content as spam or illegal.

The spam control mechanism relies on the outcome of the reputation system. It blacklists any non-legitimate content (reactive) and enslaves the publication rate of an author to the consumers satisfaction and trust in the author (proactive). This avoids flooding of unwanted content and serves as an incentive to rate to improve the quality of one’s usage experience.

We introduce the notion of a user through self-created credentials (public/private key). These guarantee the inimitability of an identity and provide authentication of already known users, non-repudiation, and integrity without the support of a central authority (CA). As these cannot prevent users from generating multiple identities and launching Sybil attacks, we minimize their effect by establishing trust in genuine identities. This is based on their social ties demonstrated by a secure pairing process and on their familiarity and similarity (community) in the underlying mobility graph.