SoD: Design of Service-based Software Systems with QoS Monitoring and Adaptation

(Start from August 2007) 

Project Summary

Service-Oriented Architecture (SOA) has the capabilities that enable the rapid composition of various services for distributed applications and the integration of the "system of systems", and hence is rapidly adopted in distributed computing systems, such as grid-enabled applications and enterprise computing infrastructures. Due to the unique characteristics of SBS, fundamental changes to current software engineering techniques are needed for designing high-quality service-based software systems (SBS), which comprise of various computing services offered by providers over networks. One major problem to achieve this goal is how to design SBS to simultaneously satisfy multiple QoS features, such as timeliness, throughput, accuracy, security, survivability, and so on. Because satisfying requirements in one QoS feature often requires sacrifices in others, tradeoffs of requirements among multiple QoS features must be analyzed and controlled in SBS. However, existing software engineering techniques cannot effectively support the analysis and adaptive control of such tradeoffs due to lack of comprehensive understanding of such tradeoffs associated with various service operations. Furthermore, SBS often operate in highly dynamic environments, where services may become temporarily unavailable due to various system and network conditions. This requires the monitoring of system situations and proper adaptation to satisfy multiple QoS requirements simultaneously.

In this multidisciplinary research proposal, the PIs from Computer Science and Engineering Department and Industrial Engineering Department with the expertise in software engineering, QoS analysis, modeling, optimization and control, software security and software simulation, propose to develop a framework effectively incorporating QoS Monitoring and Adaptation (M/A) capabilities in SBS to support the design of SBS in handling requirement tradeoffs among multiple QoS features. With the limited time and resources for the proposed project, our research will focus on four QoS features: timeliness, throughput, accuracy and security, which are important for many critical applications. To achieve this objective, our proposed research includes the following three major research tasks: (1) design of QoS monitoring capability in Adaptable SBS (ASBS), (2) design of QoS adaptation capability in ASBS, and (3) validation of design through simulations. The effectiveness of our proposed research results will be evaluated on a representative networked system testbed and a suite of simulation models for the networked system.

The intellectual merits of our proposed research include

* New scientific knowledge about the cause-effect dynamics of activity-state-event-QoS tradeoff, new analytical methodologies for discovering such knowledge from data, and a new optimization and control synthesis algorithm for QoS adaptation.

* Innovative design supports for the QoS monitoring capability, including the generation of a reliable and efficient monitoring scheme, and the automated synthesis and deployment of QoS monitoring modules.

* New simulation principles and methods enabling development of simulatable service models that can correctly interact with the emulated QoS M/A modules, and to support model configurability and scalable simulation of ASBS designs.

Our expected research results will have the following broader impacts:

a. To strengthen the scientific rigor of software design and reduce the gap between design of software systems and design of systems in engineering fields.

b. To enrich our scientific knowledge about the behavioral and performance dynamics of service operations designed primarily for functionality using logic-based operational models.

c. To be included as a part of several senior/graduate level courses on software engineering, software security, advanced information systems, and software modeling and simulation. The new material in these courses should attract high-quality students to participate in research on software design.

d. To be introduced to industry and government laboratories through the outreach programs and collaborative research activities of the Information Assurance Center at ASU and the Arizona Center for the Integrative Modeling and Simulations (ACIMS).