Less Impact Better Service (LIBS)

Abstract

We discuss a new packet service paradigm, called “Less Impact Better Service” (LIBS). In simple terms, LIBS primarily schedules packets based on the delay they cause and cancels service differentiation policies when the cumulative delay due to prioritization becomes significant for non-prioritized packets. Based on LIBS, we evaluate different service policies that prioritize small packets using different service boundaries and we show that, by and large, LIBS satisfies better a number of applications with diverse demands in delay and throughput. We emphasize on Voice over IP applications, which are delay-sensitive but also utilize small packets and rates. Among other traditional performance measures, we also measure fairness in the context of LIBS, that is, we address the question whether the delay experienced per flow is proportional to the delay caused by that flow. We obtained very promising simulation results.