Abstract:Abstract-In this paper, we present a performance model of optical burst switching (OBS) that can explain the degradation of OBS throughput performance when the control packet processing time increases. We then use the proposed performance model to investigate three feasible methods to improve OBS performance without significantly increasing the implementation complexity: addition of simple fiber delay lines (FDLs), random extra offset time, and window-based channel scheduling (WBS). Additional FDLs can elimina… Show more
“…After an offset time, the source node sends out the data burst following the same routing path of the control packet. The minimum offset time between the control packet and the data burst is Toff = H ×Tcp + Tsw, (1) whereTsw is the required switch reconfiguration time at each node, Tcp is the processing time of a control packet in a node, H is the number of hops to the destination from the current location of the control packet. Hence, H is equal to the total hop count of the path when the control packet is at the source and decreases by one for each intermediate node the control packet passes.…”
Section: Factors Affecting Throughput Degradation In Optical Burst Switchingmentioning
confidence: 99%
“…In WBS, additional Twd time will increase the equivalent control packet processing time to Tcp + Twd. We assume that Tcp is much larger than L or the Twd delay time is compensated [1]. WBS OBS assuming large Tcp does not require any additional hardware and only needs to extend the initial offset time at the source.…”
Optical Burst Switching (OBS) is a promising paradigm for high speed transmission of data. In OBS, a key problem is to schedule bursts with minimum loss. Single method is not sufficient to improve performance. So, our performance model includes some feasible methods to improve OBS performance without significantly increasing the implementation complexity. The methods are addition of simple fiber delay lines (FDLs), increasing random extra offset time, window based channel scheduling (WBS) and Burst Delay Feedback scheduling (BDFS). Additional FDLs can only eliminate the negative impact caused by the variation of the offset time between control packets and data bursts. The random extra offset time approach does not require any additional hardware in the nodes. WBS provides better throughput improvement when FDLs are used in the nodes to compensate the processing time. Finally Burst Delay Feedback Scheduling in addition with these methods can significantly improve OBS throughput and reduce transmission delay.
“…After an offset time, the source node sends out the data burst following the same routing path of the control packet. The minimum offset time between the control packet and the data burst is Toff = H ×Tcp + Tsw, (1) whereTsw is the required switch reconfiguration time at each node, Tcp is the processing time of a control packet in a node, H is the number of hops to the destination from the current location of the control packet. Hence, H is equal to the total hop count of the path when the control packet is at the source and decreases by one for each intermediate node the control packet passes.…”
Section: Factors Affecting Throughput Degradation In Optical Burst Switchingmentioning
confidence: 99%
“…In WBS, additional Twd time will increase the equivalent control packet processing time to Tcp + Twd. We assume that Tcp is much larger than L or the Twd delay time is compensated [1]. WBS OBS assuming large Tcp does not require any additional hardware and only needs to extend the initial offset time at the source.…”
Optical Burst Switching (OBS) is a promising paradigm for high speed transmission of data. In OBS, a key problem is to schedule bursts with minimum loss. Single method is not sufficient to improve performance. So, our performance model includes some feasible methods to improve OBS performance without significantly increasing the implementation complexity. The methods are addition of simple fiber delay lines (FDLs), increasing random extra offset time, window based channel scheduling (WBS) and Burst Delay Feedback scheduling (BDFS). Additional FDLs can only eliminate the negative impact caused by the variation of the offset time between control packets and data bursts. The random extra offset time approach does not require any additional hardware in the nodes. WBS provides better throughput improvement when FDLs are used in the nodes to compensate the processing time. Finally Burst Delay Feedback Scheduling in addition with these methods can significantly improve OBS throughput and reduce transmission delay.
“…We assume delay time is compensated [1]. Windowbased scheduling assuming large does not require any additional hardware and only needs to extend the initial offset time at the source.…”
Optical Burst Switching (OBS) is a promising paradigm for high speed transmission of data. In OBS, a key problem is to schedule bursts with minimum loss. Single method is not sufficient to improve performance. So, our performance model includes some feasible methods to improve OBS performance without increasing the implementation complexity. The methods include adding fiber delay lines (optical buffers), increasing offset time randomly, channel scheduling and Burst Delay Feedback scheduling (BDFS). FDLs are used only to compensate the node processing time. The random offset time approach does not require additional hardware components in the nodes. Channel scheduling in a window based manner provides better channel utilization capability when FDLs are used in the nodes. Finally Burst Delay Feedback scheduling in addition with these methods can significantly improve OBS throughput and reduce loss rate.
“…In this case, if the bandwidth requirement has been satisfied, the burst will travel through core nodes successfully; otherwise, it will be blocked. In a blocking OBS network system, various schemes have been proposed to resolve burst contention in intermediate nodes, such as optical buffering , wavelength conversion , deflection routing , and burst segmentation (BS) , to reduce the data losses in contentions.…”
Section: Introductionmentioning
confidence: 99%
“…Ana priority 0, a 1,2,3,4 = 0.8 Sim priority 0, a 1,2,3,4 = 0.8 Ana priority 0, a 1,2,3,4 = 0.6 Sim priority 0, a 1,2,3,4…”
Burst retransmission can reduce data loss in burst contention in optical burst switching (OBS) core nodes. However, uncontrolled burst retransmission may significantly increase the network load and data loss probability, defeating the purpose of retransmission. Therefore, controlled retransmission should be studied to achieve reasonable retransmission, particularly to support quality of service (QoS) in OBS networks in which priority traffic exists. In this paper, we develop a QoS-supported controlled retransmission scheme in OBS networks. Different from previous works in the literature, we set a different value for the retransmission probability at each contention for both high-and low-priority bursts, set different retransmission probabilities for bursts of different priorities for QoS support, and propose a retransmission analytical model for OBS networks. We take into account the load at each link due to both the fresh and the retransmitted traffic and calculate the path-blocking probability and the burst loss probability for high-priority and low-priority bursts to evaluate the network performance. An extensive simulation is proposed to validate our analytical model.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.