Enhancing traffic sampling scope and efficiency

Traffic Sampling is a crucial step towards scalable network measurements, enclosing manifold challenges. The wide variety of foreseeable sampling scenarios demands for a modular view of sampling components and features, grounded on a consistent architecture. Articulating the measurement scope, the required information model and the adequate sampling strategy is a major design issue for achieving an encompassing and efficient sampling solution. This is the main focus of the present work, where a layered architecture, a taxonomy of existing sampling techniques distinguishing their inner characteristics and a flexible framework able to combine these characteristics are introduced. In addition, a new multiadaptive technique proposal, based on linear prediction, allows to reduce the measurement overhead significantly, while assuring that traffic samples reflect the statistical behavior of the global traffic under analysis.