| 摘要 |
<p>In high-speed packet switching, packets of specific connections (CS1, CS2), e.g. speech, moving image, are to be handled with higher priority than packets of other connections (PS1, PS2, PS3), e.g. data, without delay jitter and without losses. …<??>In a known method, a division is made into a line-switching component (for CS1, CS2) and a packet-switching component (for PS1, PS2, PS3). The line-switching component operates using synchronous time-division multiplexing. In each case, for example, 70 packets, each comprising 40 octets, are combined on each input line to form a frame. Without special measures, this would produce delays in each switching stage of the order of 40 x 70 x 8 bits = 22,400 bits; for each input of a switching stage 22,400 bits of memory space would be required. …<??>According to the invention, all packets are divided in the exchange into packet parts of equal length and distributed among sub-frames. Switching is carried out on the basis of sub-frames. Delays and memory space requirements are reduced by the frame : sub-frame ratio. …<??>Fixed positions are preferably allocated to the higher-priority connections (CS1, CS2) in each frame and each sub-frame for the entire duration of the connection. …<??>Packet parts of the remaining packets are preferably arranged in sequence in remaining free positions and re-combined within each switching stage to form complete packets. …<IMAGE>… </p> |
