Pub Date : 2002-11-07DOI: 10.1109/HPSR.2002.1024234
Jine Chang, Chang-Soo Park
This paper proposes to reorder data bursts before allocating channels, which outperforms the conventional channel scheduling algorithms in terms of the burst loss probability, significantly reducing the voids (unused time slots). The proposed channel scheduling algorithm works better when used with a non-degenerate fiber delay lines (FDLs) buffer.
{"title":"Efficient channel-scheduling algorithm in optical burst switching architecture","authors":"Jine Chang, Chang-Soo Park","doi":"10.1109/HPSR.2002.1024234","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024234","url":null,"abstract":"This paper proposes to reorder data bursts before allocating channels, which outperforms the conventional channel scheduling algorithms in terms of the burst loss probability, significantly reducing the voids (unused time slots). The proposed channel scheduling algorithm works better when used with a non-degenerate fiber delay lines (FDLs) buffer.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124357190","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-11-07DOI: 10.1109/HPSR.2002.1024211
A. Bianco, M. Franceschinis, S. Ghisolfi, A. Hill, Emilio Leonardi, F. Neri, R. Webb
We focus on input-buffered packet switches, which are traditionally controlled slot-by-slot with heuristic approximations (such as iSLIP) of maximum weight or size matching algorithms. We consider instead a frame-based control strategy, in which f slots are grouped in a frame, and contentions are solved only at frame boundaries. A number of frame-based heuristics is described, and simulation is used to show that our approach provides performance advantages over a slot-by-slot control in a number of traffic scenarios.
{"title":"Frame-based matching algorithms for input-queued switches","authors":"A. Bianco, M. Franceschinis, S. Ghisolfi, A. Hill, Emilio Leonardi, F. Neri, R. Webb","doi":"10.1109/HPSR.2002.1024211","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024211","url":null,"abstract":"We focus on input-buffered packet switches, which are traditionally controlled slot-by-slot with heuristic approximations (such as iSLIP) of maximum weight or size matching algorithms. We consider instead a frame-based control strategy, in which f slots are grouped in a frame, and contentions are solved only at frame boundaries. A number of frame-based heuristics is described, and simulation is used to show that our approach provides performance advantages over a slot-by-slot control in a number of traffic scenarios.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124961265","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-11-07DOI: 10.1109/HPSR.2002.1024229
Jian Zhao, H. Hassanein, Jieyi Wu
Differentiated services (DiffServ) is a proposed architecture for the Internet in which various applications are supported using a simple classification scheme. Unlike its counterpart model of IntServ (integrated services) plus RSVP (resource reservation protocol), the DiffServ framework does not need to maintain large state information in core routers, and only carries out aggregate resource reservation at edge routers. Therefore, DiffServ calls for a very different routing framework from IntServ. In this paper, we propose two new QoS-based routing algorithms under the SiMO (single service, multiple options) framework for DiffServ architectures. These are kthQoSR (kth-shortest QoS routing) and EVQoSR (maximum energy value QoS routing). We also introduce QoS extensions to OSPF and compare it to our SiMO routing protocols. Our core routing strategy in SiMO, kthQoSR. is executed at the path level and selects one or more routes from a set of computed routes according to defined selection metrics. On the other hand, EVQoSR is a link level, online distributed routing algorithm. Through extensive simulation, we show that our proposed SiMO routing framework can achieve better route qualities in terms of load balancing, and network throughput.
{"title":"Quality of service based end-to-end SiMO routing framework in differentiated services networks","authors":"Jian Zhao, H. Hassanein, Jieyi Wu","doi":"10.1109/HPSR.2002.1024229","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024229","url":null,"abstract":"Differentiated services (DiffServ) is a proposed architecture for the Internet in which various applications are supported using a simple classification scheme. Unlike its counterpart model of IntServ (integrated services) plus RSVP (resource reservation protocol), the DiffServ framework does not need to maintain large state information in core routers, and only carries out aggregate resource reservation at edge routers. Therefore, DiffServ calls for a very different routing framework from IntServ. In this paper, we propose two new QoS-based routing algorithms under the SiMO (single service, multiple options) framework for DiffServ architectures. These are kthQoSR (kth-shortest QoS routing) and EVQoSR (maximum energy value QoS routing). We also introduce QoS extensions to OSPF and compare it to our SiMO routing protocols. Our core routing strategy in SiMO, kthQoSR. is executed at the path level and selects one or more routes from a set of computed routes according to defined selection metrics. On the other hand, EVQoSR is a link level, online distributed routing algorithm. Through extensive simulation, we show that our proposed SiMO routing framework can achieve better route qualities in terms of load balancing, and network throughput.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122115519","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-11-07DOI: 10.1109/HPSR.2002.1024238
Jang-Yeon Lee, Sungkwon Park
Ad hoc networks are comprised of MAC, routing protocols and physical environments. In this paper, optimum MTU (maximum transmission unit) sizes are studied in the ad hoc network environment. For performance index, "throughput rate" is defined. The throughput rate is a ratio between the number of packets received and packets sent through the overall systems. As node movements increase, losses increase because of routing route failure. MTUs between 250 bytes and 750 bytes have better throughput rates under no BER circumstances. 500 bytes MTU has the best throughput rate in 10/sup -4/ BER. AODV has a slightly better performance than DSR under no BER conditions.
{"title":"Optimum UDP packet sizes in ad hoc networks","authors":"Jang-Yeon Lee, Sungkwon Park","doi":"10.1109/HPSR.2002.1024238","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024238","url":null,"abstract":"Ad hoc networks are comprised of MAC, routing protocols and physical environments. In this paper, optimum MTU (maximum transmission unit) sizes are studied in the ad hoc network environment. For performance index, \"throughput rate\" is defined. The throughput rate is a ratio between the number of packets received and packets sent through the overall systems. As node movements increase, losses increase because of routing route failure. MTUs between 250 bytes and 750 bytes have better throughput rates under no BER circumstances. 500 bytes MTU has the best throughput rate in 10/sup -4/ BER. AODV has a slightly better performance than DSR under no BER conditions.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130395327","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-11-07DOI: 10.1109/HPSR.2002.1024636
K. Sato, K. Onda, N. Nakayama
{"title":"Performance evaluation of layer 3 switch using smartbits and evaluation of router's convergence using AX/4000","authors":"K. Sato, K. Onda, N. Nakayama","doi":"10.1109/HPSR.2002.1024636","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024636","url":null,"abstract":"","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129773575","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-11-07DOI: 10.1109/HPSR.2002.1024247
C. Zhang, M. MacGregor
Efficient fair queuing using deficit round-robin, DRR, proposed by Shreedhar and Varghese (1996) is a low-complexity packet scheduler that has several commercial implementations. DRR has also been extended as DRR+ to accommodate latency-critical flows. DRR+, however, assumes that a latency-critical flow exhibits very smooth arrivals whereas most network flows are very bursty in nature, either as the result of source bursts, or as a result of the dynamics of multihop network paths. When DRR+ encounters a burst, it reverts back to the behavior of DRR, providing no preference or latency bound for latency critical traffic. This is a fatal flaw that prevents DRR+ from being useful in scheduling bursty latency-critical flows. We present a different extension to DRR that has much lower delay and delay jitter than DRR+ and is capable of handling bursty latency-critical flows.
{"title":"Scheduling latency-critical traffic: a measurement study of DRR+ and DRR++","authors":"C. Zhang, M. MacGregor","doi":"10.1109/HPSR.2002.1024247","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024247","url":null,"abstract":"Efficient fair queuing using deficit round-robin, DRR, proposed by Shreedhar and Varghese (1996) is a low-complexity packet scheduler that has several commercial implementations. DRR has also been extended as DRR+ to accommodate latency-critical flows. DRR+, however, assumes that a latency-critical flow exhibits very smooth arrivals whereas most network flows are very bursty in nature, either as the result of source bursts, or as a result of the dynamics of multihop network paths. When DRR+ encounters a burst, it reverts back to the behavior of DRR, providing no preference or latency bound for latency critical traffic. This is a fatal flaw that prevents DRR+ from being useful in scheduling bursty latency-critical flows. We present a different extension to DRR that has much lower delay and delay jitter than DRR+ and is capable of handling bursty latency-critical flows.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134599041","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-11-07DOI: 10.1109/HPSR.2002.1024235
M. Khandker, Xiaohong Jiang, S. Horiguchi, Hong Shen
In this paper we present a new architecture for building nonblocking optical multistage interconnection networks with a given size of building block. The maximum signal loss and crosstalk of this network are much less compared to those of the widely used crossbar network with a switch complexity of O(N/sup 2/). We have found that this architecture is a good choice for constructing nonblocking optical switch networks.
{"title":"Generalized recursive network: a new architecture for nonblocking optical multistage interconnection networks","authors":"M. Khandker, Xiaohong Jiang, S. Horiguchi, Hong Shen","doi":"10.1109/HPSR.2002.1024235","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024235","url":null,"abstract":"In this paper we present a new architecture for building nonblocking optical multistage interconnection networks with a given size of building block. The maximum signal loss and crosstalk of this network are much less compared to those of the widely used crossbar network with a switch complexity of O(N/sup 2/). We have found that this architecture is a good choice for constructing nonblocking optical switch networks.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133503537","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-11-07DOI: 10.1109/HPSR.2002.1024219
S. Ioannidis, K. Anagnostakis, John Ioannidis, A. Keromytis
The ever-increasing complexity in network infrastructures is making critical the demand for network monitoring tools. While the majority of network operators rely on low-cost open-source tools based on commodity hardware and operating systems, the increasing link speeds and complexity of network monitoring applications have revealed inefficiencies in the existing software organization, which may prohibit the use of such tools in high-speed networks. Although several new architectures have been proposed to address these problems, they require significant effort in re-engineering the existing body of applications. We present an alternative approach that addresses the primary sources of inefficiency without significantly altering the software structure. Specifically, we enhance the computational model of the Berkeley packet filter (BPF) to move much of the processing associated with monitoring into the kernel, thereby removing the overhead associated with context switching between kernel and applications. The resulting packet filter, called xPF, allows new tools to be more efficiently implemented and existing tools to be easily optimized for high-speed networks. We present the design and implementation of xPF as well as several example applications that demonstrate the efficiency of our approach.
{"title":"xPF: packet filtering for low-cost network monitoring","authors":"S. Ioannidis, K. Anagnostakis, John Ioannidis, A. Keromytis","doi":"10.1109/HPSR.2002.1024219","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024219","url":null,"abstract":"The ever-increasing complexity in network infrastructures is making critical the demand for network monitoring tools. While the majority of network operators rely on low-cost open-source tools based on commodity hardware and operating systems, the increasing link speeds and complexity of network monitoring applications have revealed inefficiencies in the existing software organization, which may prohibit the use of such tools in high-speed networks. Although several new architectures have been proposed to address these problems, they require significant effort in re-engineering the existing body of applications. We present an alternative approach that addresses the primary sources of inefficiency without significantly altering the software structure. Specifically, we enhance the computational model of the Berkeley packet filter (BPF) to move much of the processing associated with monitoring into the kernel, thereby removing the overhead associated with context switching between kernel and applications. The resulting packet filter, called xPF, allows new tools to be more efficiently implemented and existing tools to be easily optimized for high-speed networks. We present the design and implementation of xPF as well as several example applications that demonstrate the efficiency of our approach.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133282988","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-11-07DOI: 10.1109/HPSR.2002.1024202
Xiaohong Jiang, M. Khandker, Hong Shen, S. Horiguchi
Vertical stacking is an interesting technique for creating nonblocking multistage interconnection networks (MINs). The rearrangeable nonblocking optical MINs are attractive since they have lower complexity than their strict nonblocking counterparts. We study the crosstalk-free permutation in rearrangeable nonblocking banyan-type optical MINs built on vertical stacking and provide a scheme for realizing crosstalk-free permutations in this class of optical MINs. The basic idea of our scheme is to decompose a permutation into multiple partial per-mutations based on the idea of Euler tour traversing, and then realize each of partial permutations crosstalk-free in a stacked copy of the MIN so that the full permutation can be realized crosstalk-free in the network in a single pass.
{"title":"Permutation in rearrangeable nonblocking optical MINs with zero first-order switching-element-crosstalk","authors":"Xiaohong Jiang, M. Khandker, Hong Shen, S. Horiguchi","doi":"10.1109/HPSR.2002.1024202","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024202","url":null,"abstract":"Vertical stacking is an interesting technique for creating nonblocking multistage interconnection networks (MINs). The rearrangeable nonblocking optical MINs are attractive since they have lower complexity than their strict nonblocking counterparts. We study the crosstalk-free permutation in rearrangeable nonblocking banyan-type optical MINs built on vertical stacking and provide a scheme for realizing crosstalk-free permutations in this class of optical MINs. The basic idea of our scheme is to decompose a permutation into multiple partial per-mutations based on the idea of Euler tour traversing, and then realize each of partial permutations crosstalk-free in a stacked copy of the MIN so that the full permutation can be realized crosstalk-free in the network in a single pass.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114370531","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2002-11-07DOI: 10.1109/HPSR.2002.1024635
A. Sasaki, H. Tsutsui, T. Saitou
{"title":"xDSL on fiber system (artfiber system)","authors":"A. Sasaki, H. Tsutsui, T. Saitou","doi":"10.1109/HPSR.2002.1024635","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024635","url":null,"abstract":"","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116203848","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}