作者: Michael Mitzenmacher , Rajmohan Rajaraman
关键词: Simulation 、 Algorithm 、 Throughput 、 Latency (engineering) 、 Cumulative distribution function 、 Computer science 、 Markov chain
摘要: Recently, several researchers have developed equations for modeling TCP behaviors, such as the expected throughput or latency, based on Markov chains derived from with additional simplifying assumptions. In this paper, we suggest new directions chain analyses of TCP. Our first contribution is to closely examine not just expectation but entire cumulative distribution function transfer times under various models. Particularly short medium transfers, likely be more useful than in terms measuring end-user satisfaction. We find that shapes functions are remarkably robust small changes model. results can extended yield approximations well expectation. Our second consider correction procedures enhance these A procedure a rule thumb allows one model used other situations. As an example, use Drop-Tail loss determine deviation between and natural models simulations. The existence simple instance suggests high-level behavior against