Implementing and evaluating network positioning in the resource addressable network

This thesis focuses on the design, implementation and evaluation of a coordinates-based mechanism to compute network positions of Internet hosts in the resource addressable network. This mechanism is based on absolute coordinates calculated from modelling the Internet as a D-dimensional Cartesian space. A small distributed set of infrastructure hosts called landmarks is chosen to first compute their coordinates and serves as a frame of reference for other hosts. I study two algorithms for landmark positioning: distributed spring algorithm and spring equilibrium algorithm. The position of the non-landmark host is calculated using a centralized algorithm. I also evaluate a schema to adjust a host's position based on the positions of its nearby peers.A location-based RAN prototype is implemented by integrating network positioning with naming and discovery modules. The prototype distribution is installed on 127 PlanetLab machines for evaluation. By performing real-time experiments on the distributed testbed, I show that the spring equilibrium algorithm outperforms the distributed spring algorithm in the aspects of efficiency and stability. Furthermore, the adjustment schema based on nearby peers is beneficial to stabilize host's position under normal network variance