| 摘要 |
A packet data communication network employs sequence numbers in message packets to identify the packets transmitted for a message, so that the order of packets can be checked, and so that the occurrence of lost or duplicated packets can be detected. A method of keeping track of these sequence numbers is provided. Any number below or above a bounded sequence number acceptance window is ignored (not accepted). Bit maps are established for the range of numbers within the bounded window, each bit representing a sequence number and the bit maps indexed by sequence number. One bit map is a received packet map, used to keep track of which sequence numbers have been received and thus to filter duplicate sequence numbers. An end-of-message bit map is used to record the positions of packets having end-of-message flags, and this is used with the received packet map to determine whether all packets of a message have been received, so that an end-of-message trigger can be generated. The sequence number and an end-of-message flag from an incoming packet are used to update these bit maps as each packet is received, then the updated bit maps used to determine the needed action. Two methods are described for tracking these sequence numbers. Both methods have good average-case performance. The first method uses an iterative loop and is more suited for software implementation. The second method uses table lookup to replace the iterative loop and is more suited for hardware implementation. This method has very good worst-case performance.
|
