Shashishekara Sitharamarao Talya

Service-based P2P overlay network for collaborative problem solving

This paper describes a service-based P2P overlay network architecture to support a collaborative environment for solving complex business processes over the network. In the proposed architecture, autonomic service providers corresponding to various activities that occur in the processes reside on the overlay network and are discovered dynamically during the execution of the process. To consummate a specific process, a set of services that map into the business process are federated together and executed in a choreographed sequence. All services have standardized interfaces and this allows any service to be seamlessly replaced with another service without affecting the performance of the federation. The paper presents two cases of application of this architecture, namely, business-to-business collaboration in an engineering environment (General Electric) and multiparty financial transactions (mortgage).

Mapping Engineering Design Processes onto a Service-grid: Turbine Design Optimization:

This study presents an application of a distributed service-oriented architecture (FIPER) for the preliminary design of gas turbines. The FIPER architecture is based on the concept of registration and discovery of services in real time. It uses a service catalog that registers the services started on the network. These services are then discovered in real time from the grid and used in the design process. The turbine preliminary design process involves changing the configuration of the turbine incrementally and evaluating its performance using analysis (simulation) code. During the design process a one-dimensional analysis code is wrapped into a service using a standard interface developed in FIPER and launched on the network. Several such services were distributed across a grid of workstations. These services were then used to support the turbine configuration optimization process. The study presents the results of the optimization as well as the scalability characteristics of the service-grid.

Constrained Optimization Using Cellular Automata

The paper presents a novel approach in understanding the topology of the search space in constrained optimization problems. The approach uses a cellular automata based methodology where the search space is represented as a grid of cells that are hypercubes in the search space. The cells coalesce together with their neighbors based on the status of the constraint violations matches in the neighboring cells. Once stabilized, the information on the topology of the search space can be used to prune the search space. It also allows designers to make more informed decisions on constraint relaxation in over-constrained problems. Techniques are suggested in the paper to accelerate the process of coalescing and to reduce the computational complexity of the process. Several standard mathematical functions from the literature have been used to demonstrate this methodology. The paper describes the turbine preliminary design optimization in which the objective is to maximize the turbine efficiency with constraints on aerodynamic parameters. In addition, it describes the application of the proposed technique to the problem. A cellular automaton is a discrete dynamical system consisting of identical elements whose behavior is completely specified in terms of local rules. They consist of infinite number of cells arranged in a regular grid such that each cell can exist in one of the finite number of possible states. The grid can be two-dimensional consisting of cells or multi-dimensional where it consists of hypercubes. Each cell in the grid is aware of the state of its neighboring cells and changes its own state based on the states of its neighbors at any instance of time. By repeating such interactions with neighbors, a collective behavior pattern of the entire grid emerges. In cellular automata, all the objects are assembled from the same kind of structural elements and are subject to the same laws. These laws or rules are uniformly applied to all the cells to determine their state at each time step. The paper proposes an optimization methodology that uses cellular automata by describing local rules in a search space grid to cluster the search space based on the pattern of constraint violations and to obtain the optima within each of the clusters. Clustering is defined as an attempt to find natural groups of components (or data) based on some similarity metric. Clustering algorithms have been employed in the past to group large sets of data into clusters of similar data. Often, the distance between two points is taken as a common metric to assess the similarity among population components. The proposed algorithm clusters the data based on the pattern of constraint violations. There are two