Today's Internet is under enormous pressure: there is a growing diversity of use models, an urgent need for trustworthy communication, and a growing set of stakeholders who must coordinate to provide Internet services. The project is developing the eXpressive Internet Architecture (XIA), a potential new network architecture that is built around several key principles. First, XIA represents a single network that offers inherent support for communication between diverse principals including hosts, content, services, and unknown future entities. For each type of principal, XIA defines a narrow waist that dictates the API for communication and the network communication mechanisms. Second, XIA rests upon a foundation of intrinsic security in which the integrity and authenticity of communication is guaranteed using XIA's various self-certifying identifiers. Third, XIA enables flexible context-dependent mechanisms for establishing trust between the communicating principals, bridging the gap between human and intrinsically secure identifiers. Finally, recognizing the Internet's central role in our society, economic, policy, and usability considerations play a major role in XIA's design. The project includes user experiments to evaluate and refine the interface between the network and users, and studies that analyze the relationship between technical design decisions, and economic incentives and public policy. The outcome of the project will be knowledge about a potential future network architecture that is inherently trustworthy, supports long-term evolution of network use models and network technology, and provides explicit interfaces for interaction between network actors. The architecture will also enable greater visibility and control for various stakeholders, including users, ISPs, and content owners.
We designed and implemented a future Internet architecture named XIA that is trustworthy, supports long-term evolution of usage models, supports long-term technology evolution, and supports explicit and well-defined interfaces between network actors. In so doing, this future design remediates some key limitations of the existing TCP/IP architecture, notably the absence of built-in security, and limited ability for the architectural components themselves to evolve. The realization of this architecture is embodied by the Linux XIA implementation, implemented and publicly available on the GitHub open-source repository https://github.com/AltraMayor/XIA-for-Linux/. Major outcomes of the award are presented in the major publications associated with this award, in HotNets '11, NSDI '12, and CCR '14, as well as in the two BU Ph.D. dissertations produced from this award: Chong Wang (2012) and Michel Machado (2014). One major milestone was the Linux kernel implementation of XIA, which, by the end of the award, was complete with a functional network stack, including XDP/XIP, which are clean-slate replacements for UDP/IP, and several new principal types. A second major milestone is the demonstration of evolvability within XIA, as embodied by the porting of other future Internet components from other projects, and thus never designed to interoperate with XIA, onto the Linux XIA substrate. A notable example of this, achieved in the final year of the award, is the completed port of the Serval service-centric architecture to provide a service identifier (SID) implementation within XIA.
