Secure in-network big data provision with suspension chain model

Big data raises strong demands on a network infrastructure to support the efficient data provision and retrieval. Information-Centric Networking (ICN) approach is an emerging approach to satisfy this demand, where big data is ubiquitously cached at the intermediate physical entities (IPEs) in the network and retrieved through names. However, the unpredictability of users, IPEs, copy holders, and publishers makes the network suffer from malicious-request attack and data-poisoning attack. To inhibit such attacks, we design a trust model, namely a suspension-chain model (SCM), which is a trust chain that seamlessly merges certificate authority (CA)-based trust and neighbor-based trust. Based on SCM, we propose a data-centric authentication mechanism (DCAuth) integrating certificate collection and packet forwarding, where the suspension certificate chain can be constructed for any authentication to the unpredictable users/IPEs/publishers without accessing servers. We conduct simulations to evaluate the performance of the DCAuth, which show that malicious-request and data-poisoning attacks can be efficiently prevented by the DCAuth.