| Related sites for http://www.cs.bgu.ac.il/~discsp/ |
| Robots_net__Is_Sony\'s_OPEN-R_Really_Open? OPEN-R has been touted by Sony for nearly two years as an open standard for robotics development however information from Sony on OPEN-R, other than press releases, is difficult to find. (April | | RankHigher_Search_EngineTraffic_Booster Works by creating large amounts of "doorway" pages which all point to the website's main page. Up to 7000 pages may be created at the click of a button. | | SVS_Index Public access systems, hardware, NetShift software and services. | | Earthcam_Surfbot Free tool for reviewing Earthcam's list of online webcams. Enables the viewer to selectively filter out any gender and present the results in ascending or descending order by viewer count. | | Allied_Networks_Inc_ Computer and network connectivity products, gigabit ethernet cables, copper and fiber optics. | | RFC_3015 Megaco Protocol Version 1.0. F. Cuervo, N. Greene, A. Rayhan, C. Huitema, B. Rosen, J. Segers. November 2000. | | NetContinuum,_Inc_ ASIC-based web application firewall products and services. Company and product information, application security threat information, press coverage and contacts. | | CPU_Specific_Languages__Assembly_Languages Link collection on assemblers, simulators, interpreters, all with source code. [RocketAware.com] | | Paul_Thurrott\'s_SuperSite_for_Windows__Windows_XP_Tablet_PC_Edition Features a review of the Tablet PC operating system with screenshots. | | Diemen_Repository_of_Interaction_Design_Patterns Open source (Wiki) collection and collaboration site focused on creating and extending patterns for screen design, with a focus on Interaction Design and Usability. | | MidiJoiner Chains together MIDI and KAR files into a single file. | | TIVR_Communications Offer a H.264 video decoder which support real time decoding up to QVGA resolution on mobile devices. [Series 60] | | xn6_net Offers free URL redirection. | | Basic_Web_Design General web design information. | | Trademark_Protection_on_the_Internet_in_for_Non_Top-Level_Domains_Used_as_Top-level_Domains Overview of trademark protection for country code domain names marketed as alternatives to top level domains. | | Simple_XML_Data_Manipulation_Language Consists of a data definition and manipulation language like SQL and a database API. | | Universal_Parcel_Shipping_Software Packaging and Shipping software exclusively designed for the retail shipping business. Program supports all major carriers worldwide. Built-in entire UPS and FedEx Worldwide Country/Zone database. [Wi | | Dean_Systems Offers design, e-commerce, and search engine optimization services. Located in Denver, Colorado, United States. | | Lizard_King_Arts Scenic programs including sunsets, wildflowers, and wildlife. | | D++CMS A module based web content management framework (PHP). |
|
DisCSP Home Embedded menu. Please use modern browser! Distributed CSP Home Computer Science Dept., Ben-Gurion University Welcome to the Distributed Constraint Satisfaction home, at Ben-Gurion University in the Negev, Israel. This is also the home of our research group — however, we strive to make it useful to the general DisCSP community, so your comments are welcome. If you object to your paper appearing here (perhaps because it is preliminary), please tell me about that. Distributed Constraint Satisfaction Problems In Distributed CSPs, a set of variables is distributed among agents. The variables are connected by constraints which define the constraints network among the agents. As a result, the search algorithm for solving these problems is a distributed algorithm. The algorithm is run by agents that communicate by sending and receiving messages. In general, messages contain information about assignments of values to variables and refutations of assignments, by agents that have no compatible assignment, to their own variables. The first to propose this family of problems was Makoto Yokoo, who published two distributed search algorithms for solving DisCSPs — Asynchronous Backtracking (Yokoo et al., 1992) Asynchronous Weak-Commitment Search (Yokoo, 1995) The Asynchronous Backtracking algorithm was the base for later and clearer versions published by Yokoo (1998, 2000). Versions of the asynchronous backtracking algorithm, which include Dynamic Backtracking methods and do not change the initial constraints network, were published by Bessière et. al. in 2001. A paper draft on the advantages and drawbacks of the different versions of asynchronous BT is to be published in 2003. Because of the distributed nature of DisCSPs, the solving algorithm must run on all agents and perform a complete and correct search, by exchanging messages among the agents. This generates multiple research questions: Should the algorithm be asynchronous or synchronous? What are good measures of performance for distributed search algorithms? What family of distributed CSPs should serve for experimental evaluations of algorithms? Can distributed search benefit by using analogies to standard search on CSPs, like ordering variables and values, etc.? What everyday distributed combinatorial search problems can be formulated as DisCSPs? All of the above issues have been tackled in the last 4 years of intensive research on DisCSPs and on distributed search algorithms for DisCSPs. However, the answers to issues of DisCSP algorithm design and performance measurements are only preliminary. In fact, papers to date differ widely by their experimental methods and by their most basic results about algorithms performance.... Several real-world problems that can be represented as a distributed constraint satisfaction problem have been published in recent years. These problems are distributed by nature and cannot be solved with centralized methods. SensorDCSP, a problem of multiple sensors and multiple targets was defined by DARPA and was at the focus of several papers on DisCSPs (Fernandez et al., 2002). Recent papers referred to the Meeting scheduling problem in which agents try to agree on a time and place for meetings between them. A large family of DisCSPs, distributed problems of timetabling is all around us — from scheduling multiple departments in a University while sharing lecture halls, to scheduling big projects of CERN. Other applications involved problems of communication networks... Research directions The simplest form of a distributed search algorithm for DisCSPs could be described as follows: Order all agents. Let agent #1 (A1) perform an assignment to its variable. Next, agent A1 sends the assignment on a message to agent A2. Every agent awaits a message from the agent before it. When agent Ai receives the message, it attempts to assign its variable, such that all constraints with former assigned variables are satisfied. If a compatible assignment is found, agent Ai adds it to the message and sends the message agent Ai+1. If not, it sends the same message back to agent Ai-1. A solution is reached when the last agent (An) finds an assignment for its variable. A no-solution is declared after the first agent (A1) receives a backtrack message on its last value, i.e., its domain empties. The above algorithm is usually called Synchronous Backtracking (SBT). It is clear that no two agents are performing computations at the same time in SBT. It is also correct, by a clear analogy to chronological backtrack. This naive algorithm is a base point for all distributed algorithms. The main drawback Yokoo was seeking to improve was the idleness of agents through most of the search. He has done so by allowing agents to perform computations asynchronously. But idleness of agents is not the major flaw in SBT. By using BackJumping we can improve immensely the performance of SBT though the algorithm stays synchronous. SBT defines a base line from which algorithms which exploit the asynchronous nature of the problems are developed. One approach, asynchronous search, was investigated extensively during the last decade (Yokoo, Bessière, Meseguer, Silaghi, Hamadi). Other approaches, which use some coordination among agents in order to explore different parts of the search space concurrently, or to use one variable ordering according to a selected heuristic, and to avoid computation against inconsistent partial assignment were suggested, tested, and found very effective, by members of our group. Evaluating performance of solving algorithms for DisCSPs is a major issue which is investigated by our group. The asynchronous, multi-machine environment in which the algorithms run, makes it difficult to measure the efficiency of the algorithm in terms of time. Furthermore, time is not the only relevant measure for distributed algorithms. The amount of messages the network is loaded with and the concurrent computational effort are just as relevant. Our algorithmic papers reflect on proper experimental evaluation methods for comparing distributed search algorithms on DisCSPs, and the results include several measures and behavior under message delays.  Maintained by Alon Grubshtein CSP Constraint satisfaction is a general problem in which the goal is to find values for a set of variables that will satisfy a given set of constraints. It is the core of many applications in artificial intelligence, and has found its application in many areas, such as planning and scheduling. DisCSP In distributed constraint satisfaction, variables, constraints or both are distributed among a set of independent but communicating agents. |
|
