HRT provides data interpretation

HRT provides the data interpretation

On assumed that the re transmission is non-persistent, IE, once entered in queuing system data a collision, the involved stations re transmit their messages after a random time adjournment following an exponential law. As the population is infinite messages, each message can be treated as if it queue management system station belongs to a new queuing system data . This means that each message new arrival can find an empty station. The system can be considered if there is an infinite number of stations. One can show that Λ is Poissonian (a Poissonian flux is added and explodes “is split”).

The operation of the non-persistent CSMA protocol segmented is the same as that of non-segmented CSMA except in that the first l time axis is segmented into “slots” of duration τ (maximum propagation delay between any two stations). It is assumed that frame size is a constant which is a multiple m of τ. A frame arriving at a queue management system station, before being issued, must wait for the start of the “slot” next and that the station has detected a queuing system data medium to silence the start time of the “slot”. The transmission time of a frame therefore lasts m/ τ slots (an integer). If the medium is sensed busy, the issue will be postponed to a random delay (non-persistent).

As the study of non-segmented CSMA assumes that the number of stations is infinite and that the entire input stream (new+ frames re transmission collisions in frames) forms a Poissonian process Λ rates. The figure below shows what can be observed on the medium. On can see that idle period lasts at least one “slot” and a transmission period successfully or unsuccessfully last m+ τ seconds. A collision occurs when two or more frames arrive within one “slot” and are transmitted at the beginning of “slot” pro chain. L'etude a real system is seldom feasible in an environment operational.

The representation of the operation of a system of more or less accurately queue management system is necessary to allow us to approach the system behavior. This representation is called modélisation. Extant given to study a system, we must first create a model of it. It should be stressed that such a model is always a compromise between accuracy and simplicity. Indeed queue management system there is no single model for a given real system.

According to the knowledge of each system, we can get either a very loyal but often not workable model, slightly faithful and very workable model. Our task here is to find a model that is relatively faithful to the real and workable system. Once the model is obtained queue management system the next step is to conduct a study (often statistical in the case of a discrete event system) characterization applications supported by the system. This characterization is often results in a random or deterministic stream description (material, information).
Big image
Big image