GridNets 2006 – October 1 st Grid Resource Management by means of Ant Colony Optimization Gustavo Sousa Pavani and Helio Waldman Optical Networking Laboratory.

Slides:



Advertisements
Similar presentations
Ch. 12 Routing in Switched Networks
Advertisements

Mobile Ad-hoc Network Simulator: mobile AntNet R. Hekmat * (CACTUS TermiNet - TU Delft/EWI/NAS) and Radovan Milosevic (MSc student) Mobile Ad-hoc networks.
Ch. 12 Routing in Switched Networks Routing in Packet Switched Networks Routing Algorithm Requirements –Correctness –Simplicity –Robustness--the.
Data and Computer Communications
Hadi Goudarzi and Massoud Pedram
SDN + Storage.
A Centralized Scheduling Algorithm based on Multi-path Routing in WiMax Mesh Network Yang Cao, Zhimin Liu and Yi Yang International Conference on Wireless.
Security Issues in Ant Routing Weilin Zhong. Outline Swarm Intelligence AntNet Routing Algorithm Security Issues in AntNet Possible Solutions.
An Energy Efficient Routing Protocol for Cluster-Based Wireless Sensor Networks Using Ant Colony Optimization Ali-Asghar Salehpour, Babak Mirmobin, Ali.
Data and Computer Communications Ninth Edition by William Stallings Chapter 12 – Routing in Switched Data Networks Data and Computer Communications, Ninth.
Graded Channel Reservation with Path Switching in Ultra High Capacity Networks Reuven Cohen, Niloofar Fazlollahi, David Starobinski ECE Dept., Boston University.
The Antnet Routing Algorithm - A Modified Version Firat Tekiner, Z. Ghassemlooy Optical Communications Research Group, The University of Northumbria, Newcastle.
By Libo Song and David F. Kotz Computer Science,Dartmouth College.
Routing Strategies Fixed Routing
Beneficial Caching in Mobile Ad Hoc Networks Bin Tang, Samir Das, Himanshu Gupta Computer Science Department Stony Brook University.
Mobile Agents for Adaptive Routing Presented by Hong-Jiun Chen & Manu Prasanna Gianni Di Caro & Marco Dorigo.
December 20, 2004MPLS: TE and Restoration1 MPLS: Traffic Engineering and Restoration Routing Zartash Afzal Uzmi Computer Science and Engineering Lahore.
Investigation of antnet routing algorithm by employing multiple ant colonies for packet switched networks to overcome the stagnation problem Firat Tekiner.
Performance Analysis of Wavelength-Routed Optical Networks with Connection Request Retrials Fei Xue+, S. J. Ben Yoo+, Hiroyuki Yokoyama*, and Yukio Horiuchi*
Ant Colonies As Logistic Processes Optimizers
In-Band Flow Establishment for End-to-End QoS in RDRN Saravanan Radhakrishnan.
Ants-based Routing Marc Heissenbüttel University of Berne
TUDelft Knowledge Based Systems Group Zuidplantsoen BZ Delft, The Netherlands Roland van der Put Léon Rothkrantz Routing in packet switched networks.
Swarm Intelligent Networking Martin Roth Cornell University Wednesday, April 23, 2003.
Jennifer Rexford Princeton University MW 11:00am-12:20pm Wide-Area Traffic Management COS 597E: Software Defined Networking.
1 Algorithms for Bandwidth Efficient Multicast Routing in Multi-channel Multi-radio Wireless Mesh Networks Hoang Lan Nguyen and Uyen Trang Nguyen Presenter:
Algorithms for Self-Organization and Adaptive Service Placement in Dynamic Distributed Systems Artur Andrzejak, Sven Graupner,Vadim Kotov, Holger Trinks.
The Research of Applying Random Early Blocking strategy to Dynamic Lightpath Routing National Yunlin University of Science & Technology.
AntNet: Distributed Stigmetric Control for Communications Networks Gianni Di Caro & Marco Dorigo Journal of Artificial Intelligence Research 1998 Presentation.
Distributed Quality-of-Service Routing of Best Constrained Shortest Paths. Abdelhamid MELLOUK, Said HOCEINI, Farid BAGUENINE, Mustapha CHEURFA Computers.
IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS 2007 (TPDS 2007)
1 Pertemuan 20 Teknik Routing Matakuliah: H0174/Jaringan Komputer Tahun: 2006 Versi: 1/0.
1 Introducing Routing 1. Dynamic routing - information is learned from other routers, and routing protocols adjust routes automatically. 2. Static routing.
M.Menelaou CCNA2 ROUTING. M.Menelaou ROUTING Routing is the process that a router uses to forward packets toward the destination network. A router makes.
A Dynamically Adaptive Hybrid Algorithm for Scheduling Lightpaths in Lambda-Grids Neena R. Kaushik and Silvia M. Figueira Santa Clara University Presented.
NTU GICE Swarm Intelligence for Routing in Communication Networks Speaker: Shih-Chun Lin Advisor: Kwang-Cheng Chen.
Stochastic sleep scheduling (SSS) for large scale wireless sensor networks Yaxiong Zhao Jie Wu Computer and Information Sciences Temple University.
“Intra-Network Routing Scheme using Mobile Agents” by Ajay L. Thakur.
Algorithms for Allocating Wavelength Converters in All-Optical Networks Authors: Goaxi Xiao and Yiu-Wing Leung Presented by: Douglas L. Potts CEG 790 Summer.
CSCI 465 D ata Communications and Networks Lecture 15 Martin van Bommel CSCI 465 Data Communications & Networks 1.
Load-Balancing Routing in Multichannel Hybrid Wireless Networks With Single Network Interface So, J.; Vaidya, N. H.; Vehicular Technology, IEEE Transactions.
Data Communications and Networking Chapter 11 Routing in Switched Networks References: Book Chapters 12.1, 12.3 Data and Computer Communications, 8th edition.
Connectivity-Aware Routing (CAR) in Vehicular Ad Hoc Networks Valery Naumov & Thomas R. Gross ETH Zurich, Switzerland IEEE INFOCOM 2007.
Presentation of Wireless sensor network A New Energy Aware Routing Protocol for Wireless Multimedia Sensor Networks Supporting QoS 王 文 毅
Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.
GPSR: Greedy Perimeter Stateless Routing for Wireless Networks EECS 600 Advanced Network Research, Spring 2005 Shudong Jin February 14, 2005.
1 st EELA Grid School Distributed Simulation of Multiple Failure Events on Optical Networks Gustavo S. Pavani 1, Nelson T. Yunaka 2, Tatiana C. Figueiredo.
Vishal Jain, AntNet Agent Based Strategy for CMDR “Agent Based Multiple Destination Routing ”
Routing Networks and Protocols Prepared by: TGK First Prepared on: Last Modified on: Quality checked by: Copyright 2009 Asia Pacific Institute of Information.
1 Presented by Sarbagya Buddhacharya. 2 Increasing bandwidth demand in telecommunication networks is satisfied by WDM networks. Dimensioning of WDM networks.
AntNet: A nature inspired routing algorithm
November 4, 2003Applied Research Laboratory, Washington University in St. Louis APOC 2003 Wuhan, China Cost Efficient Routing in Ad Hoc Mobile Wireless.
1 Dynamic RWA Connection requests arrive sequentially. Setup a lightpath when a connection request arrives and teardown the lightpath when a connection.
M ulti m edia c omputing laboratory Biologically Inspired Cooperative Routing for Wireless Mobile Sensor Networks S. S. Iyengar, Hsiao-Chun Wu, N. Balakrishnan,
Dynamic Control of Coding for Progressive Packet Arrivals in DTNs.
Data Consolidation: A Task Scheduling and Data Migration Technique for Grid Networks Author: P. Kokkinos, K. Christodoulopoulos, A. Kretsis, and E. Varvarigos.
Using Ant Agents to Combine Reactive and Proactive strategies for Routing in Mobile Ad Hoc Networks Fredrick Ducatelle, Gianni di caro, and Luca Maria.
Routing in Delay Tolerant Network Qing Ye EDIFY Group of Lehigh University.
Peter Pham and Sylvie Perreau, IEEE 2002 Mobile and Wireless Communications Network Multi-Path Routing Protocol with Load Balancing Policy in Mobile Ad.
Distance Vector Routing
1 Traffic Engineering By Kavitha Ganapa. 2 Introduction Traffic engineering is concerned with the issue of performance evaluation and optimization of.
Performance Comparison of Ad Hoc Network Routing Protocols Presented by Venkata Suresh Tamminiedi Computer Science Department Georgia State University.
The Multi-agent System for Dynamic Network Routing Ryokichi Onishi The Univ. of Tokyo, Japan.
Name : Mamatha J M Seminar guide: Mr. Kemparaju. GRID COMPUTING.
William Stallings Data and Computer Communications
Authors: Jiang Xie, Ian F. Akyildiz
Isabella Cerutti, Andrea Fumagalli, Sonal Sheth
Lecture XVII: Distributed Systems Algorithms Inspired by Biology
Firat Tekiner (Phd Student) Z. Ghassemlooy
Chapter-5 Traffic Engineering.
Presentation transcript:

GridNets 2006 – October 1 st Grid Resource Management by means of Ant Colony Optimization Gustavo Sousa Pavani and Helio Waldman Optical Networking Laboratory – OptiNet Communications Department – Decom School of Electrical and Computer Engineering – FEEC State University of Campinas – UNICAMP.

GridNets 2006 – October 1 st Gustavo Sousa Pavani 2 Introduction Resource management in grids is a very challenging task, since the resources to be shared are very heterogeneous: CPU time, storage space or data, and network bandwidth Need for networks that are dynamically configurable in order to support large transfers of data – Use of a GMPLS control plane for managing network resources Use of the Ant Colony Optimization meta-heuristics to tackle the problem of grid scheduling, by the adaptation of the AntNet framework, which is a ACO-based algorithm, to this problem – Use of ant-like mobile agents to act on behalf of the user to make scheduling decisions – Considered the allocation of processing power and provisioning of lightpaths

GridNets 2006 – October 1 st Gustavo Sousa Pavani 3 Proposed Algorithm – ACO ACO is used to refer to the class of algorithms that are inspired in the process of foraging for food by natural ants for the optimization of hard-to-solve problems It is characterized by ant-like mobile agents that cooperate and stochastically explore a network, iteratively building solutions based on their own memory and on the traces (pheromone levels) left by other agents The AntNet framework is an ACO-based algorithm used for routing of packets in telecommunication networks – The original AntNet framework uses the delay introduced by each hop as the metric for routing. This is not applicable in circuit-switching networks, where the main metric is generally the number of hops. Indeed, the minimization of the number of hops of a connection following some criteria is a very good heuristics to reduce the overall blocking probability in a network

GridNets 2006 – October 1 st Gustavo Sousa Pavani 4 Proposed Algorithm – Outline At regular intervals, a forward ant is launched from a random source node to another random destination node. In its trip, the forward ant will select the next hop i using a random scheme that accounts the path selection probabilities, given by the pheromone levels in each neighbor link, and a heuristics value, calculated from the congestion of each neighbor links During its trip, the forward ant gathers the label of each node where it passes by, putting it in its memory. When it arrives at the destination node, it becomes a backward ant, collects the availability information and returns to the source node using the same path followed by the forward ant, but in the opposite direction. At each intermediate node i, it updates the routing table, comparing its memory with the local parametric model If the ant does not reach its destination node in a number of pre-established hops it is dropped. This avoids lost ants circulating forever in the network In a hop-by-hop basis, the lightpath is routed evaluating the entry of the pheromone routing table that matches the destination. The neighbor with the higher level of pheromone is chosen as the next hop

GridNets 2006 – October 1 st Gustavo Sousa Pavani 5 Proposed Algorithm – Adaptations to AntNet Data Structures 1 At each node i, we have the following data structures: – Pheromone-routing table T i : It is a matrix containing a row for each destination of the network and a column for each neighbor node, for storing the pheromone values. The sum of each row must be equal to 1 Example of the pheromone routing table of node 4: Destinations Neighbor Nodes

GridNets 2006 – October 1 st Gustavo Sousa Pavani 6 Proposed Algorithm – Adaptations to AntNet Data Structures 2 – Statistical parametric model M i : It is a matrix containing the triplet for each destination d of the network. These values are updated using an exponential model within the non-sliding window of w observations Example of the statistical parametric model of node 4: dd dd EdEd Destinations Neighbor Nodes

GridNets 2006 – October 1 st Gustavo Sousa Pavani 7 Proposed Algorithm – Adaptations to AntNet Data Structures 3 – Availability vector A i : It is a vector containing an availability metric for each destination of the network. The rationale behind this structure is to allow for the load-balancing of the server farms Example of the availability vector of node 4: Availability Destinations

GridNets 2006 – October 1 st Gustavo Sousa Pavani 8 Proposed Algorithm – Routing of Forward Ants Congestion level for each neighbor link: Probability of a forward ant to choose the neighbor node: –  gives the emphasis between pheromone level (long-term memory) and instantaneous congestion state (short-term memory)

GridNets 2006 – October 1 st Gustavo Sousa Pavani 9 Proposed Algorithm – Updating by Backward Ants 1 At each intermediate node i through source node, the backward ant updates the parametric model M i and routing table T i of the intermediate nodes k  [( i +1), d [, i.e., it treats these nodes as destinations. However, this is done only if the evaluation of sub-path traced by the ant is good enough, i.e., its number of hops is less than the superior estimate I sup. This allows for the updating of good paths found by ants that are not intended to those destinations. The update is always done for the destination node d at i

GridNets 2006 – October 1 st Gustavo Sousa Pavani 10 Proposed Algorithm – Updating by Backward Ants 2 First, the parametric model M i k is updated. After, an adaptive reinforcement r is calculated for updating the routing table: The obtained r is limited to 0.9 to avoid stagnation and its value is “squashed” Now, if the neighbor node m is on the path, then it receives a positive reinforcement: The other nodes receives a negative reinforcement:

GridNets 2006 – October 1 st Gustavo Sousa Pavani 11 Proposed Algorithm – Loop Avoidance and Management Memory of the forward ant works as tabu list for selecting the next hop Dead-end: all neighbors of the node where the forward ant is being processed are already visited – Ignore the heuristic correction given by the congestion information Use only pheromone values – Remove all nodes that belongs to the cycle from the ant’s memory

GridNets 2006 – October 1 st Gustavo Sousa Pavani 12 Proposed Algorithm – Path Computation A B C D PATH B D C C D C Calculated hop-by-hop!

GridNets 2006 – October 1 st Gustavo Sousa Pavani 13 NSFNet Network 14 nodes and 21 bi-directional links Node 1 as user, other nodes as resources

GridNets 2006 – October 1 st Gustavo Sousa Pavani 14 Simulation Homogeneous Poissonian traffic with uniform spatial profile – Node 1 as the only source of lightpath requests Fixed duration of the lightpath: 14.4 sec (equivalent to a 1.8 GB file transfer at a 10 Gbps rate) Fixed processing time of each job: 12 hours – Homogeneous processors – Same number of processors per node First-fit wavelength assignment algorithm Blocking can be caused by lack of resources on the: – Optical network – Server farm

GridNets 2006 – October 1 st Gustavo Sousa Pavani 15 Results – Blocking Probability

GridNets 2006 – October 1 st Gustavo Sousa Pavani 16 Results – Total Grid Workload W = 8 Load = 4 Erlangs The optical network can limit the total grid workload (W = 2 and 4) for higher loads

GridNets 2006 – October 1 st Gustavo Sousa Pavani 17 Results – The Effect of the Ant Launching Rate The increase on the ant launching rate can enhance the system performance in terms of blocking probability

GridNets 2006 – October 1 st Gustavo Sousa Pavani 18 Results – The Average Overhead on Control Channels due to the Ants

GridNets 2006 – October 1 st Gustavo Sousa Pavani 19 Conclusions Alternative for grid scheduling when the optical network resources have to be dynamically controlled. Integrated management of processing and network resources. Load balancing of the: – Processing workload – Optical network Use of a GMPLS control plane for lightpath provisioning – Advantages of the proposed algorithm over link-state protocols Some scenarios proposed where the performance of the grid was limited by either networking resources or processing power

GridNets 2006 – October 1 st Gustavo Sousa Pavani 20 Any questions?