Erdoğan Doğdu
Research
| Web Services |
|---|
Web Services Compositions and Efficient Scheduling Strategies Web Services compositions provide new services by combining a number of web services calls in a transactional specification. Languages like BPEL are the specification languages used in creating web services compositions. Composition languages are currently in development and they only deal with static services and their composition. Future web services compositions need to be more dynamic in which service compositions should be able to provide dynamically integrated services in execution time. Therefore, future web services execution engines for web services compositions need to incorporate techniques for finding and integrating web services in run-time. We developed simple scheduling techniques for selecting the best available services so that the overall system performance is high. |
 |
Funding:
|
| Bioinformatics |
An Integrated Bioinformatics System for Collaboration Using Web Services There are many publicly available biological data stores on the web that allow researchers to search and download bioinformatics data. Most of these systems require human interaction through a web browser where data is manually entered for processing. Integration of these databases is expected to be a difficult problem for near future. One of the goals of our system will be the integration with other databases and systems in that searching multiple remote databases instantly with limited user intervention will be possible. This requires utilization of latest distributed system integration technologies, or in other words "middleware", such as web services and standard Internet protocols for communiication. By doing so, we will achieve an "interoperable" system. One of the benefits of using standard web protocols and the latest web services middleware is to be able to interconnect heterogeneous hardware and software systems, or in other words, "interoperability". For example, our initial system will be implemented on a Sun Microsystems hardware running Solaris operating system and using Java programming language. Yet, it should be able to interact with an online bioinformatics database implemented on a Windows system developed using .NET. Funding:
|
Middleware |
SyD: System of Mobile Devices Middleware I have contributed to the design and development of a mobile application development middleware called "System of Mobile Devices", or SyD in short, developed at Georgia State University. SyD is developed to overcome the difficulties in developing mobile distributed applications over small heterogeneous devices such as PDAs, cellphones, etc. SyD provides ready-to-use modules via an application programming interface (API) towards this goal. Following are the services SyD handles:
|
 |
Funding:
|
Real-Time Database Systems |
|---|
Real-Time Database Transactions with Execution Histories Developed efficient scheduling strategies for processing real-time database transactions with execution histories. In this framework, transaction execution times are logged when transactions are repetitevly executed. This information is utilized when ordering the scheduling of these transactions to achieve a "fair scheduling" outcome in which transactions of diffent classes in size are scheduled fairly equally (with similar completion ratios). Classical transaction scheduling techniques for real-time database systems, such as the Earliest Deadline First priority assignment technique, schedules transactions unfairly, in the sense, short transactions have better completion ratios within their deadlines. Via a detailed computer simulation system, we have tested and shown that the techniques we developed overcomes this problem. Funding:
|
Estimation of Aggregate Queries in Real-Time Databases Developed a "double sampling" method to estimate the result of COUNT queries in real-time database systems with given error quotas. In this system, user specifies the degree of correctness of the estimation by an error quota. System adjusts the sampling size of the estimation technique for varying error quotas and accordingly tries to find the best estimate within given time constraints. Funding:
|
