| 摘要 |
The invention relates to a method for implementing buffering in a packet-switched telecommunications network. In the method, incoming data units are received and stored in a logical queue in a buffer (BF) comprising a plurality of memory locations, and data units are read out from the memory location (FML) at the head of the queue at a predetermined rate. In order to eliminate delay variations over desired connections by as simple a method as possible, the traffic is divided into at least two different classes in such a way that (a) the data units of one class are stored in the first free memory location starting from the head of the queue in each case, (b) feed points (FP; FP1 ... FP3) are provided for the other classes at predetermined points in the middle of the queue formed by the memory locations, one for each of the other classes, and (c) a data unit belonging to one of said other classes is stored in the feed point for said class when the memory location associated with said feed point is free, and in the first free memory location after the feed point in the queue when the memory location associated with said feed point is reserved. |
