Towards verifiable parallel content retrieval

Many software vendors are providing mechanisms for parallel content retrieval using multiple connections, e.g., parallel HTTP channels, to increase the availability and reliability of the download procedure. At the same time, there is no native verification mechanism to support simultaneous content verification from multiple sources. While it is possible to set-up multiple TLS tunnels with different sources, there is no guarantee that the data itself is authentic since the trust is placed on the connection and not in the data itself. In this paper we present a parallel verification mechanism based on hash trees, allowing clients to authenticate data segments regardless of the container from where they were retrieved from, but just with the original provider. Some benefits of the proposed mechanism include low CPU processing costs, low verification overhead and possibility to support legacy data.