Pub Date : 2002-08-07DOI: 10.1109/ISPAN.2002.1004304
Xiaohong Jiang, Hong Shen, M. Khandker, S. Horiguchi
Vertical stacking is an alternative for constructing nonblocking multistage interconnection networks (MINs). In this paper, we study the crosstalk-free permutation in rearrangeable, self-routing Banyan-type optical MINs built on vertical stacking and propose a new scheme for realizing permutations in this class of optical MINs, crosstalk-free. The basic idea of the new scheme is to classify permutations into permutation classes such that all permutations in one class share the same crosstalk-free decomposition pattern. By running the Euler-Split based crosstalk-free decomposition only once for a permutation class and applying the obtained crosstalk-free decomposition pattern to all permutations in the class, crosstalk-free decomposition of permutations can be realized in a more efficient way. We show that the number of permutations in a permutation class is huge, enabling the average time complexity of the new scheme to realize a crosstalk-free permutation in an N by N network to be reduced to O(N) from previously O(NlogN).
{"title":"A new scheme to realize crosstalk-free permutations in optical MINs with vertical stacking","authors":"Xiaohong Jiang, Hong Shen, M. Khandker, S. Horiguchi","doi":"10.1109/ISPAN.2002.1004304","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004304","url":null,"abstract":"Vertical stacking is an alternative for constructing nonblocking multistage interconnection networks (MINs). In this paper, we study the crosstalk-free permutation in rearrangeable, self-routing Banyan-type optical MINs built on vertical stacking and propose a new scheme for realizing permutations in this class of optical MINs, crosstalk-free. The basic idea of the new scheme is to classify permutations into permutation classes such that all permutations in one class share the same crosstalk-free decomposition pattern. By running the Euler-Split based crosstalk-free decomposition only once for a permutation class and applying the obtained crosstalk-free decomposition pattern to all permutations in the class, crosstalk-free decomposition of permutations can be realized in a more efficient way. We show that the number of permutations in a permutation class is huge, enabling the average time complexity of the new scheme to realize a crosstalk-free permutation in an N by N network to be reduced to O(N) from previously O(NlogN).","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132467204","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 : 2002-08-07DOI: 10.1109/ISPAN.2002.1004302
C. Lam, F. Lau
We present an improved version of a distance-vector protocol for routing in networks with unidirectional links, which has the added capabilities of handling link and node faults as well as addition and deletion of nodes. Because of the unidirectional nature of links, the design of a correct and efficient protocol for such networks is nontrivial. We discuss the behavior and performance of the proposed protocol.
{"title":"Fault-tolerant routing algorithms for unidirectional networks","authors":"C. Lam, F. Lau","doi":"10.1109/ISPAN.2002.1004302","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004302","url":null,"abstract":"We present an improved version of a distance-vector protocol for routing in networks with unidirectional links, which has the added capabilities of handling link and node faults as well as addition and deletion of nodes. Because of the unidirectional nature of links, the design of a correct and efficient protocol for such networks is nontrivial. We discuss the behavior and performance of the proposed protocol.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128237659","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 : 2002-08-07DOI: 10.1109/ISPAN.2002.1004271
Rafael P. Saldaña, Winfer C. Tabares, W. E. Yu
Four basic cellular automata (CA) algorithms are implemented on a Beowulf cluster with 8 processors using MPI and C. The CA algorithms are, namely, (1) Game of Life, (2) Greenburg-Hasting, (3) Cyclic Space, and (4) Hodgepodge Machine. The mathematical and computational aspects of 1D and 2D CA are also considered.
{"title":"Parallel implementations of cellular automata algorithms on the AGILA high performance computing system","authors":"Rafael P. Saldaña, Winfer C. Tabares, W. E. Yu","doi":"10.1109/ISPAN.2002.1004271","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004271","url":null,"abstract":"Four basic cellular automata (CA) algorithms are implemented on a Beowulf cluster with 8 processors using MPI and C. The CA algorithms are, namely, (1) Game of Life, (2) Greenburg-Hasting, (3) Cyclic Space, and (4) Hodgepodge Machine. The mathematical and computational aspects of 1D and 2D CA are also considered.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117331234","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 : 2002-08-07DOI: 10.1109/ISPAN.2002.1004286
Jaime D. L. Caro, Tedros Weldemicael Zeratsion
The generalized de Bruijn digraph denoted by G/sub B/(n, m) is defined to be the digraph with m vertices labelled by 0, 1, 2, ..., m-1 and with the adjacency defined as follows: If i is a vertex in G/sub B/(n, m) then i is connected to each vertex in the set E(i), where E(i)={ni+/spl alpha/(mod m)|/spl alpha//spl isin/[0, n-1]}. The generalized de Bruijn graph denoted by UG/sub B/(n, m) is defined to be the undirected version of G/sub B/(n, m) obtained by replacing each arc by an undirected edge and eliminating self-loops and multi-edges. In this paper we show that the diameter of UG/sub B/(n, m) is 2 for any m in [n+1, n/sup 2/] where n divides m and that the diameter is 3 for any m in [n/sup 2/+1, n/sup 3/] where n divides m.
{"title":"On the diameter of a class of the generalized de Bruijn graphs","authors":"Jaime D. L. Caro, Tedros Weldemicael Zeratsion","doi":"10.1109/ISPAN.2002.1004286","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004286","url":null,"abstract":"The generalized de Bruijn digraph denoted by G/sub B/(n, m) is defined to be the digraph with m vertices labelled by 0, 1, 2, ..., m-1 and with the adjacency defined as follows: If i is a vertex in G/sub B/(n, m) then i is connected to each vertex in the set E(i), where E(i)={ni+/spl alpha/(mod m)|/spl alpha//spl isin/[0, n-1]}. The generalized de Bruijn graph denoted by UG/sub B/(n, m) is defined to be the undirected version of G/sub B/(n, m) obtained by replacing each arc by an undirected edge and eliminating self-loops and multi-edges. In this paper we show that the diameter of UG/sub B/(n, m) is 2 for any m in [n+1, n/sup 2/] where n divides m and that the diameter is 3 for any m in [n/sup 2/+1, n/sup 3/] where n divides m.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129495738","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 : 2002-05-22DOI: 10.1109/ISPAN.2002.1004258
Adora E. Calaor, A. Hermosilla, Bobby O. Corpus
In this study, a solution to the school timetabling problem using a parallel genetic algorithm with simulated annealing is presented. The hybridization of simulated annealing and the parallel genetic algorithm is explained. Running these algorithms in parallel on a local network of workstations is also discussed. Comparative results among the different parallel models are given. Implementation of the parallel algorithms is used to construct conflict-free and satisfactory timetables for the Department of Mathematics of the University of the Philippines Diliman. The program output of this study can be easily modified to be used as a helpful and efficient guide in the decision-making process of the scheduler.
{"title":"Parallel hybrid adventures with simulated annealing and genetic algorithms","authors":"Adora E. Calaor, A. Hermosilla, Bobby O. Corpus","doi":"10.1109/ISPAN.2002.1004258","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004258","url":null,"abstract":"In this study, a solution to the school timetabling problem using a parallel genetic algorithm with simulated annealing is presented. The hybridization of simulated annealing and the parallel genetic algorithm is explained. Running these algorithms in parallel on a local network of workstations is also discussed. Comparative results among the different parallel models are given. Implementation of the parallel algorithms is used to construct conflict-free and satisfactory timetables for the Department of Mathematics of the University of the Philippines Diliman. The program output of this study can be easily modified to be used as a helpful and efficient guide in the decision-making process of the scheduler.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128791317","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 : 2002-05-22DOI: 10.1109/ISPAN.2002.1004278
M. Justan, F. P. Muga, I. H. Sudborough
In this study, we are particularly interested in one class of symmetric interconnection networks, namely the pancake graph, P/sub n/. Pancake graphs are especially attractive for distributed processing because they compare favorably with a hypercube of similar size. They have smaller degree and diameter than correspondingly large hypercubes. The one-sided pancake graph, P/sub n/, has been modeled as a Cayley graph on S/sub n/, the symmetric group of order n, while the two-sided pancake graph, 2P/sub n/, has been represented as a Cayley graph on the wreath product S/sub 2//spl bsol/S/sub n/. In this paper, we want to generalize the pancake graph, i.e., state the m-sided pancake flipping problem, and describe its graph as the m-sided pancake graph, mP/sub n/. Specifically, 1. we shall model the m-sided pancake graph, mP/sub n/ as Cayley graph on the wreath product of some finite groups; then, 2. we shall look into the degree, diameter and the routing protocol of the m-sided pancake graph, mP/sub n/; 3. we shall give the bounds for the diameters of mP/sub n/; and, 4. we shall give the diameters of 3P/sub n/ for 1 /spl les/ n /spl les/ 6.
{"title":"On the generalization of the pancake network","authors":"M. Justan, F. P. Muga, I. H. Sudborough","doi":"10.1109/ISPAN.2002.1004278","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004278","url":null,"abstract":"In this study, we are particularly interested in one class of symmetric interconnection networks, namely the pancake graph, P/sub n/. Pancake graphs are especially attractive for distributed processing because they compare favorably with a hypercube of similar size. They have smaller degree and diameter than correspondingly large hypercubes. The one-sided pancake graph, P/sub n/, has been modeled as a Cayley graph on S/sub n/, the symmetric group of order n, while the two-sided pancake graph, 2P/sub n/, has been represented as a Cayley graph on the wreath product S/sub 2//spl bsol/S/sub n/. In this paper, we want to generalize the pancake graph, i.e., state the m-sided pancake flipping problem, and describe its graph as the m-sided pancake graph, mP/sub n/. Specifically, 1. we shall model the m-sided pancake graph, mP/sub n/ as Cayley graph on the wreath product of some finite groups; then, 2. we shall look into the degree, diameter and the routing protocol of the m-sided pancake graph, mP/sub n/; 3. we shall give the bounds for the diameters of mP/sub n/; and, 4. we shall give the diameters of 3P/sub n/ for 1 /spl les/ n /spl les/ 6.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"148 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115878920","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 : 2002-05-22DOI: 10.1109/ISPAN.2002.1004295
W. Bein, L. Larmore, Charles O. Shields, I. H. Sudborough
We describe total congestion 1 embeddings of complete binary trees into three dimensional grids with a fixed number of layers. More specifically, we give a one-to-one embedding of any complete binary tree into a hexahedron shaped grid such that no tree nodes or edges occupy the same grid positions. With 7 layers, the number of nodes in the grid is at most 1.09375 times the number of nodes in the tree and with 5 layers we obtain a ratio of 75/64=1.171875. Unlike more standard embeddings these embeddings intricately weave the branches of various subtrees into each other. Finally using a standard recursive method, for 2 layers a ratio of 39/32=1.21875 can be obtained.
{"title":"Fixed layer embeddings of binary trees","authors":"W. Bein, L. Larmore, Charles O. Shields, I. H. Sudborough","doi":"10.1109/ISPAN.2002.1004295","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004295","url":null,"abstract":"We describe total congestion 1 embeddings of complete binary trees into three dimensional grids with a fixed number of layers. More specifically, we give a one-to-one embedding of any complete binary tree into a hexahedron shaped grid such that no tree nodes or edges occupy the same grid positions. With 7 layers, the number of nodes in the grid is at most 1.09375 times the number of nodes in the tree and with 5 layers we obtain a ratio of 75/64=1.171875. Unlike more standard embeddings these embeddings intricately weave the branches of various subtrees into each other. Finally using a standard recursive method, for 2 layers a ratio of 39/32=1.21875 can be obtained.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127629604","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 : 2002-05-22DOI: 10.1109/ISPAN.2002.1004284
Christine Lagman, P. Tagle
The authors propose an improvement to the interconnected parallel banyan network (IPBN), a highspeed switching architecture for converged services networks. The IPBN design consists of multiple banyan networks with links provided at every stage to allow cell transfer to and from each plane, thereby offering multiple paths. The improvement makes use of an aggressive transmission scheme, which takes advantage of this multiplane architecture by sending out multiple copies of each cell. Using simulations, aggressive transmission in the IPBN is shown to offer better performance by way of lowering cell loss probabilities. Furthermore, the introduction of a skewing function in the distribution of cell copies to the IPBN's multiple planes resulted in even better performance in the way of further decreased cell loss probabilities.
{"title":"Aggressive transmission in the interconnected parallel banyan network","authors":"Christine Lagman, P. Tagle","doi":"10.1109/ISPAN.2002.1004284","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004284","url":null,"abstract":"The authors propose an improvement to the interconnected parallel banyan network (IPBN), a highspeed switching architecture for converged services networks. The IPBN design consists of multiple banyan networks with links provided at every stage to allow cell transfer to and from each plane, thereby offering multiple paths. The improvement makes use of an aggressive transmission scheme, which takes advantage of this multiplane architecture by sending out multiple copies of each cell. Using simulations, aggressive transmission in the IPBN is shown to offer better performance by way of lowering cell loss probabilities. Furthermore, the introduction of a skewing function in the distribution of cell copies to the IPBN's multiple planes resulted in even better performance in the way of further decreased cell loss probabilities.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126991021","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 : 2002-05-22DOI: 10.1109/ISPAN.2002.1004287
Cheng-Wei Chen, Chung-Kai Chen, Jenq-Kuen Lee
Proposes an ontology specification for JavaBeans programs (JavaBeans being the object component model of Java). Our specification is written using the DAML+OIL (DARPA Agent Markup Language + Ontology Interchange Language) language, which is based on the RDF (Resource Description Framework) schema and the XML syntax. The vocabulary of this ontology provides a basic terminology to annotate components with information about conditions and suggestions for adopting a component for specialization. It also gives a reference criterion for choosing the most suitable components at a given time, and an environment for performance and functionality purposes. With our design of the annotations, it's also possible to automatically retrieve the annotations of object components, connect them by their object-oriented relationships, organize them to form component databases and discover them in the databases by using the component characteristics. This facilitates the sharing of component resources on the Internet. In addition, we also give an application scenario for employing this ontology specification.
{"title":"Building ontology for optimization and composition of parallel JavaBean programs","authors":"Cheng-Wei Chen, Chung-Kai Chen, Jenq-Kuen Lee","doi":"10.1109/ISPAN.2002.1004287","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004287","url":null,"abstract":"Proposes an ontology specification for JavaBeans programs (JavaBeans being the object component model of Java). Our specification is written using the DAML+OIL (DARPA Agent Markup Language + Ontology Interchange Language) language, which is based on the RDF (Resource Description Framework) schema and the XML syntax. The vocabulary of this ontology provides a basic terminology to annotate components with information about conditions and suggestions for adopting a component for specialization. It also gives a reference criterion for choosing the most suitable components at a given time, and an environment for performance and functionality purposes. With our design of the annotations, it's also possible to automatically retrieve the annotations of object components, connect them by their object-oriented relationships, organize them to form component databases and discover them in the databases by using the component characteristics. This facilitates the sharing of component resources on the Internet. In addition, we also give an application scenario for employing this ontology specification.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121133725","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 : 2002-05-22DOI: 10.1109/ISPAN.2002.1004272
Daniel G. Pamintuan, E. Albacea
In this paper, we introduce DISTREP, a multiplatform and operating system independent file distribution and replication package. It features dynamic distribution and low-overhead network connections that efficiently conserve network bandwidth. A complete replication to M machines of N bytes in DISTREP may only require a transfer of at least N bytes, less than the NM bytes that other systems require. The system also offers easy addition of compression methods and other bandwidth-limiting algorithms (like the ability to integrate rsync into the system) to further improve and speed up file transfers.
{"title":"File replication and distribution system for low bandwidth networks","authors":"Daniel G. Pamintuan, E. Albacea","doi":"10.1109/ISPAN.2002.1004272","DOIUrl":"https://doi.org/10.1109/ISPAN.2002.1004272","url":null,"abstract":"In this paper, we introduce DISTREP, a multiplatform and operating system independent file distribution and replication package. It features dynamic distribution and low-overhead network connections that efficiently conserve network bandwidth. A complete replication to M machines of N bytes in DISTREP may only require a transfer of at least N bytes, less than the NM bytes that other systems require. The system also offers easy addition of compression methods and other bandwidth-limiting algorithms (like the ability to integrate rsync into the system) to further improve and speed up file transfers.","PeriodicalId":255069,"journal":{"name":"Proceedings International Symposium on Parallel Architectures, Algorithms and Networks. I-SPAN'02","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130156317","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：