Osher Yadgar

Scaling-Up Distributed Sensor Networks: Cooperative Large-Scale Mobile-Agent Organizations

We present a system called the Distributed Dispatcher Manager (DDM) for effectively managing very large-scale networks of thousands of sensor agents and thousands of objects. DDM makes use of a hierarchical team organization in which the solution process is distributed into smaller fragments of problems that can be solved partially by simple agents. We present extensive experimental results which indicate that problems involving hundreds and thousands of Dopplers and targets cannot be solved in a traditional flatarchitecture. We also present a new sensor tracking algorithm through which a single agent can track an object by taking multiple sequential measurements and combining them. We then suggest ways to combine partial solutions to form a global solution. We show that the number of levels in the hierarchy influences the accuracy of results. As the number of levels increases the number of tracked targets drops, even though this drop is moderate. However, as the number of levels increases the time every agent needs to complete its mission drops exponentially. By combining these two results DDM can achieve a balance between these two properties.

Goal Satisfaction in Large Scale Agent-Systems: A Transportation Example

We demonstrate the applicability of a low complexity physics-oriented approach to a large-scale transportation problem. The framework is based on modeling cooperative MAS by a physics-oriented model. According to the model, agent systems inherit physical properties, and therefore the evolution of the computational systems is similar to the evolution of physical systems. We provide a detailed algorithm to be used by a single agent and implement this algorithm in our simulations. Via these we demonstrate effective task allocation and execution in an open, dynamic MAS that consists of thousands of agents and tasks.

Hierarchical information combination in large-scale multiagent resource management

In this paper, we describe the Distributed Dispatcher Manager (DDM), a system for managing resource in very large-scale task and resource domains. In DDM, resources are modeled as cooperative mobile teams of agents and objects or tasks are assumed to be distributed over a virtual space. Each agent has direct access to only local and partial information about its immediate surroundings. DDM organizes teams hierarchically and addresses two important issues that are prerequisites for success in such domains: (i) how agents can extend local, partial information to arrive at a better local assessment of the situation and (ii) how the local assessments from teams of many agents can be integrated to form a global assessment of the situation. We conducted a large number of experiments in simulation and demonstrated the advantages of the DDM over other architectures in terms of accuracy and reduced inter-agent communication.

Hierarchical organizations for real-time large-scale task and team environments

In this paper, we describe the Distributed Dispatcher Manager (DDM), a system for monitoring large collections of dynamically changing tasks. We assume that tasks are distributed over a virtual space. Teams consist of very large groups of cooperative mobile agents. Each agent has direct access to only local and partial information about its immediate surroundings. DDM organizes teams hierarchically and addresses two important issues that are prerequisites for success in such domains: (i) how agents should process local information to provide a partial solution to nearby tasks, and (ii) how partial solutions should be integrated into a global solution. We conducted a large number of experiments in simulation and demonstrated the advantages of the DDM over other architectures in terms of accuracy and reduced inter-agent communication * .

From Local Search to Global Behavior: Ad Hoc Network Example

We introduce the Consensual N-Player Prisoners Dilemmaas a large-scale dilemma. We then present a framework for cooperative consensus formation in large-scale MAS under the N-Person Prisoners Dilemma. Forming consensus is performed by demonstrating the applicability of a low-complexity physics-oriented approach to a large-scale ad hoc network problem. The framework is based on modeling cooperative MAS by a physics percolation theory. According to the model, agent-systems inherit physical properties, and therefore the evolution of the computational systems is similar to the evolution of physical systems. Specifically, we focus on the percolation theory, the emergence of self-organized criticality, and the exploitation of phase transitions. We provide a detailed low-ordered algorithm to be used by a single agent and implement this algorithm in our simulations. Via these approaches we demonstrate effective message delivery in a large-scale ad hoc network that consists of thousands of agents.

Emergent ad hoc sensor network connectivity in large-scale disaster zones

We introduce a disaster-zone-monitoring web-based application. While using this application, the user may simulate different large-scale disaster zones. He may use tools to define a disaster zone and its communication requirements. The web application uses the Google Earth infrastructure and is publicly available to use online during the conference from every computer connected to the Internet. The evolving deployment of nodes will be updated each second to reflect the current state of the entities residing within the monitored zone. The user will be able to navigate through the disaster zone to inspect the dynamically changed environment, and to learn about node deployment and the current network connectivity and service availability. He could ask the system to find the number of agents required to be deployed and present this deployment. The user also will be able to define the budget limitations. Thus, the system will derive the number of agents, their deployment, and the resulting system utility. Given a system utility, the user can decide whether to adopt the deployment even though it does not guarantee full coverage, or to increase the budget to improve system utility.

Coverage density as a dominant property of large-scale sensor networks

Large-scale sensor networks are becoming more present in our life then ever. Such an environment could be a cellular network, an array of fire detection sensors, an array of solar receptors, and so on. As technology advances, opportunities arise to form large-scale cooperative systems in order to solve larger problems in an efficient way. As more large-scale systems are developed, there is a growing need to (i) measure the hardness of a given large-scale sensor network problem, (ii) compare a given system to other large-scale sensor networks in order to extract a suitable solution, (iii) predict the performance of the solution, and (iv) derive the value of each system property from the desired performance of the solution, the problem constraints, and the users preferences. The following research proposes a novel system term, the coverage density, to define the hardness of a large-scale sensor network. This term can be used to compare two instances of large-scale sensor networks in order to find the suitable solutions for a given problem. Given a coverage density of a system, one may predict the solution performance and use it jointly with the preference and the constraints to derive the value of the systems properties.

Airspace management of autonomous UAVs

One major issue currently preventing the adoption of autonomous unmanned air vehicles (UAVs) is the lack of airspace management to prevent the UAVs from colliding with each other, with human-piloted planes or helicopters, with static objects such as buildings, and with dynamic flying objects such as flocks of birds. In this work, we present a novel airspace management approach to autonomous UAVs. Our airspace management system allows UAVs to dynamically and autonomously choose between three modes of operation: (i) centralized, (ii) cooperative decentralized, (iii) noncooperative decentralized.