Automatic originator regulation of IMS multiple traffic by stateless signaling prioritization

Priority-Based Queue Management Scheme to Reduce Overloads on SIP Servers

“…In RED, packets are dropped based on a dropping probability that depends on the average queue length. In contrast to the original RED proposed in [7], which did not distinguish packet types, our proposed scheme for a SIP server enables the calculation of different dropping probabilities according to the SIP method type because [5] and [6] showed that queue management based on giving a priority to each SIP method type could improve throughput under overload. The proposed scheme can calculate different dropping probabilities according to the SIP method type even if the queue length is the same when receiving a new SIP message.…”

“…This results in higher throughput in the network when overload occurs. W. Zhu et al pointed out that the scheme proposed in [6] is inefficient due to its use of two queues. They provided an example of this inefficiency in [6] as follows.…”

“…W. Zhu et al pointed out that the scheme proposed in [6] is inefficient due to its use of two queues. They provided an example of this inefficiency in [6] as follows. When the high-priority queue is full and the low-priority queue is not, the received non-INVITE messages must be dropped even if there is plenty of space in the low-priority queue.…”

“…The proposed scheme adopts a single-queue model to deal with the question of queue utilization efficiency discussed in [6]. It is also based on RED (Random Early Detection), which is a queue management scheme for a router [7].…”

“…The queue management scheme proposed in [6], however, needs to search the queue at every reception of a new SIP message when its queue is full. Assuming that SIP servers receive a large number of packets in an overload situation, the use of their resources on searching the queue may cause further overload.…”