ABSTRACT

With continued miniaturization and increasing computation power, computing systems are becoming deeply embedded into ordinary life and interact with physical processes and events. They monitor the physical world with sensors and provide appropriate reaction and control over it. The scale of such cyber-physical interactions is very wide; embedded systems of such interactions range from resource-constrained stand-alone devices to global-scale networked embedded systems. This cyber-physical interaction, which occurs through ubiquitous embedded systems, has the potential to transform how humans interact with and control the physical world. Some applications of such systems include agile manufacturing, military applications, advanced traffic control, environmental control, and electric power grid control. For many of these applications, providing a real-time database (RTDB) is essential since they are inherently data-intensive. Without RTDB support, data should be managed manually, increasing the difficulty of the application development. Databases of such applications need to process transactions within their deadlines using fresh temporal data representing the current real world status, e.g., the current power grid or battlefield status. Several RTDBs have been proposed. However those solutions have not fully considered the unique challenges that arise at different scales of embedded systems. A list of the challenges includes (1) resource impoverishment, (2) distinctive H/W resources, (3) openness, (4) large scale, and (5) interaction with physical world. The primary goal of this proposal is to address the aforementioned challenges in providing QoS guarantees in RTDBs for networked embedded systems. We propose a QoS-aware real-time database for networked embedded systems. We take a step-wise approach to address the problem. First, we focus on a QoS guarantee problem in a RTDB for a stand-alone embedded system. Here, the major research issue is to deal with resource constraints, distinctive H/W resources, and unpredictable workloads. Next, we move on to guaranteeing QoS when stand-alone embedded systems are networked and coordinated in wide-area open environments. The solutions for stand-alone embedded systems are extended to address additional challenges such as long remote data access delays, large-scale, and complex interactions between distributed embedded systems. Our approach is based on the extensive application of feedback control theory to provide systematic QoS management and predictable real-time data services. This approach is especially effective when the workload behavior is difficult to model in a predictable way. In our proposed approach, the key QoS requirement for a RTDB is the timeliness of transactions, and it can be met in a cost-effective manner through the adaptive QoS management scheme. The expected contributions in the proposed QoS-aware RTDB for networked embedded systems consist of (1) QoS adaptation technique in resource-constrained stand-alone real-time embedded systems, (2) decentralized QoS management technique in distributed real-time embedded systems, (3) implementation of the proposed QoS management scheme for broader impact to the research community, and (4) demonstration of the overall improved performance. Other Recent and Upcoming Colloquia