Alex Bordetsky

Mobile system for precise aero delivery with global reach network capability

This paper discusses the current status of the development of the mobile aerial delivery system to be further employed in a variety of different applications. High accuracy of the developed system enables its use in precision troop resupply, precise sensors placement, urban warfare reconnaissance and other similar operations. This paper overviews the overall system architecture and components of the developed aero delivery system itself and then proceeds with describing the current status of integrating it with an advanced deployment platform, unmanned aerial system, to achieve mobility and autonomy of operations. The paper also discusses some other systems in development pursuing similar goals and reviews some novel applications that become possible with the developed aerial delivery system.1

Agent-based support for collaborative data mining in systems management

Addresses the issues of structuring and supporting the collaborative data mining process. This paper extends the technology of multi-participant decision-making support into the data mining process and describes a perspective agent-based support architecture for data mining committees. The proposed architecture is based on agent-facilitators integrated with the layered structure of a case-based reasoning memory and an artificial neural network to support collaborative classification and linkage analysis between the heterogeneous views of system managers. A telecommunications management example is used to illustrate the approach.

Multicriteria analysis tools in real-life problems

Applied optimization problems such as design, identification, design of controlled systems, operational development of prototypes, analysis of large-scale systems, and forecasting from observational data are multicriteria problems in essence. Construction of the feasible solution set is of primary importance in the above problems. The definition of a feasible solution set is usually considered to be the skill of a designer. Even though this skill is essential, it is by no means sufficient for the correct statement of the problem. There are many antagonistic performance criteria and all kinds of constraints in these problems; therefore, it is quite difficult to correctly determine the feasible set. As a result, ill-posed problems are solved, and optimal solutions are searched for far from where they should be. As a consequence, the optimization results have no practical meaning. In this work we propose methods and tools that will assist the designer in defining the feasible solution set correctly.

Intercepting a Target with Sensor Swarms

This paper introduces a new coordination method to intercept a mobile target in urban areas with a team of sensor platforms. The task is to intercept the target before it leaves the area. The approach combines algorithmic concepts from ant colony and particle swarm optimization in order to bias the search and to spread the team in the search area. The algorithms introduced are tested in simulation experiments on grids. The success probabilities measured are relatively high for most parameter combinations, and the target is intercepted in roughly half the simulation time on average. Furthermore, the experiments reveal robust behavior with regard to the parameter setting.

Uncooperative multi-agent communication network control, hybrid LQ approach

Retaining connectivity of a mesh network of noncooperative multiple mobile agents is considered. The new connectability theory is applied to preserving network connectivity of non-cooperative mobile wireless networks. The newly introduced connectability matrix allows detecting a loss of network connectivity, as well as determining which nodes are isolated. The hybrid system techniques are used to describe the network dynamics, and LQR is used to control the motion of the relay agents in the communication network with changing dimension. A numerical example demonstrates the efficacy of the proposed techniques.

Multi-criteria approach in configuration of energy efficient sensor networks

Problem and Motivation: A designing of energy efficient wireless sensor networks is one of the most trendy research topics. In this paper the authors developed a new multi-criteria approach which allows filtering out non-efficient topologies during a phase of designing of a sensor network and also running a self-adjustment process during its functioning. It is shown that the problem of designing of an energy efficient wireless sensor network belongs to the class D of optimization combinatorial problems for which the equivalent recognition problems are NP-complete or open. The D-class (Peltsverger B., Khavronin O. (1999)) is such that according to a decomposition approach there exist particular criteria compatible with the main one, and the recognition task corresponding to a particular criterion always belongs to the P-class problem.

Adaptive Management of QoS Requirements for Wireless Multimedia Communications

We present a control model, which provides response time and bandwidth requirement adaptation in audio, video, and application sharing multipoint IP teleconferences for emerging wireless multimedia communications. The model is based on revealing feedback controls for multimedia call preparation and subsequent real time connection control. Case-based reasoning memory is used to associate real time congestion (connection) controls with call preparation controls and user QoS profiles. Web agents are used to capture user and application multimedia call profiles observed at the application layer and transfer them into the case memory. RTP statistics are used to identify the connection management feedback controls for the network layer. Real-time adaptation at the network layer and above is made possible by using hierarchical coding techniques. The proposed adaptive management architecture is illustrated by a case memory representation of call preparation feedback controls, RTP feedback control tests for providing audio stream bandwidth adaptation, and configuration of integrated experiments.

A Feedback Control Model for Managing Quality of Service in Multimedia Communications

Advances in Internet Protocol (IP) multicasting provide a rich background for support of IP multipoint collaborative communications. IP multicast applications have traditionally been deployed over wired networks, however, new wireless technologies such as Local Multipoint Distribution Service (LMDS) are currently emerging as transport mechanisms for wireless video, voice, and application sharing environments. For multimedia wireless services to effectively evolve, management tools that can support Quality of Service (QoS) adaptation of increasingly complex network resources and customer application profiles are needed. In this paper, we present a control model which provides response time and bandwidth requirement adaptation of audio, video, and application sharing multipoint IP teleconferences for emerging wireless multimedia communications. Our study is innovative in that it integrates feedback controls between the application and network layers. Our model is based on revealing feedback controls for multimedia call preparation and subsequent real-time connection control. Case-based reasoning memory is used to match real-time congestion (connection) controls with call preparation controls and user profiles for improved QoS. Network agents are used to capture user and multimedia teleconference application call profiles at the application layer and transfer them to the case memory. Real Time Protocol (RTP) statistics are used to identify connection management feedback controls at the network layer. Receiver-based, real-time adaptation at the network layer and above is possible through the use of a hierarchical coding technique. The proposed adaptive management architecture is based upon a case memory representation of call preparation feedback controls, RTP feedback controls for providing audio stream bandwidth adaptation, and configuration descriptions for integrated experiments. We conclude that implementation of these techniques should lead to improved QoS of wireless IP multipoint teleconferences.

Predictive modeling and retaining connectivity for mobile wireless communication network

The problem of maintaining connectivity in a mobile communication network is considered. The agents consist of two classes: one class of agents, which is termed primary mission agents, performs surveillance operations that may or may not be coordinated efforts. The second class of agents, which is termed relay agents, is controlled to maintain the connectivity. Two control algorithms are proposed to retain the connectivity of the network. Sliding mode observers are employed to predict the trajectories of the primary mission agents, when only their positions are known from blue force messages, and this information is used to construct state dependent graphs encapsulating the measure of connectivity. An optimal model predictive control is employed to develop the control policies for the relay agents to maximize connectivity. Also, the hybrid system approach is used to retain the connectivity of the communication network. The efficacy of the proposed control algorithms is confirmed on a case study of Riverine detection and interdiction operation.

Knowledge management for wireless grid operation enters

Our objective is to develop generalized principles for grid operations centers that can be applied to specific domains. We focus on the domain of wireless telecommunications NOCs (WNOCs) as a starting point in this process. To address this problem, we focus specifically on the knowledge requirements for service-based decision support within wireless telecommunications. We use a Mission-Policy-Metrics-Sensor management control feedback model and associated influence diagrams to show how an integrated NOC can be developed from the constituent wireless NOCs (WNOCs). We derive metrics for the various WNOCs, which, in turn, drive data, model, user interface, and agent requirements for a decision environment that can support integrated customer service. By concentrating on a knowledge-based view of WNOCs, we provide a methodology for integrating existing grid-based operations centers.