| 摘要 |
Method for queueing packets, each packet of said packets comprising timing information representing a remaining time until a deadline associated with delivery at its destination; said system comprising: N queues (511 to 515), each queue thereof being configured for buffering at least one packet, said N queues having a cyclic order; said method comprising: keeping track of a pointer (503), initially pointing to a queue of said N queues; afterwards, repeatedly incrementing said pointer over said N queues, according to said cyclic order, after each passing of a scheduling interval S; enqueueing each packet of said packets in a queue of said N queues, that queue differing from the queue to which said pointer is pointing, based on said packet's timing information and taking into account said cyclic order; and dequeueing enqueued packets from said N queues, while prioritizing dequeueing from the queue to which said pointer is pointing. This helps to prevent that an urgent packet arriving in a busy queue to which the pointer is pointing might get stuck in that queue after the pointer increments to the next queue. The queuing system (103) comprises and enqueuing module (502) for enqeuing (503) each packet of said packets (501) in a queue of said plurality of N queues (511 to 515). A dequeuing module dequeues (504) the enqueued packets from the queues (511 to 512), while prioritizing dequeuing from the queue (511) to which said pointer is pointing. Preferably, each packet's remaining time falls within a pre-determined permissible time range R, and said scheduling interval S represents an amount of time essentially equal to said time range R divided by N-1. In this manner, the permissible time range for the packets can be covered effectively by the queues excluding the queue to which the pointer is pointing. Explicit timing information in packets may be used to indicate a deadline associated with the packet, i.e., by when a packet needs to be delivered to the destination. |
