Bikash Sabata

Taxonomy for QoS specifications

It is becoming increasingly commonplace for multiple applications with different quality of service (QoS) requirements to share the resources of a distributed system. Within this environment, the resource management algorithms must take into account the QoS desired by applications and the ability of the system resources to provide it. We present a taxonomy for specifying QoS for the different components of a distributed system, from the applications down to the resources. We specify QoS as a combination of metrics and policies. QoS metrics are used to specify performance parameters, security requirements and the relative importance of the work in the system. We define three types of QoS performance parameters: timeliness, precision, and accuracy. QoS policies capture application-specific policies that govern how an application is treated by the resource manager. Examples of such policies are management policies and the levels of service. We explore each of these components of the QoS taxonomy in detail.

Modeling applications for adaptive QoS-based resource management

The paper describes two innovative models that facilitate adaptive QoS driven resource management in distributed systems comprising heterogeneous computing, storage, and communication resources. The first model, denoted the Logical Application Stream Model (LASM), recursively captures a distributed applications structure, resource requirements, and relevant end to end quality of service (QoS) parameters. Upon invocation of the application by a user, the resource manager can use this model to initially structure the end to end application, allocate resources to this application, and schedule this application on these resources, so as to provide QoS to all applications and to efficiently utilize system resources. Later, when the system state changes, the resource manager can use this application model to dynamically reallocate, reschedule, and restructure applications. The recursive nature of the model enables application developers to easily model large scale applications. We also describe a model, denoted the Benefit Function (BF), that captures user QoS preferences and enables the resource manager to gracefully degrade application QoS under certain conditions.

Surface Correspondence and Motion Computation from a Pair of Range Images

The estimation of the motion transformation of a moving object from a sequence of images is of prime interest in computer vision. In this paper, the issues in estimating the motion parameters from a pair of range images are addressed. The motion estimation task, in the domain of range image sequences, has two components: (1) extract the surfaces and establish the correspondence of the surfaces over the frames in the sequence of range images, and (2) compute the motion transformation using these surface correspondences. A novel procedure based on a hypergraph representation is presented for finding surface correspondence. Two scenes are modeled as hypergraphs and the hyperedges are matched using a subgraph isomorphism algorithm. The hierarchical representation of hypergraphs not only reduces the search space significantly but also facilitates the encoding of the topological and geometrical information used to direct the search procedure. Results obtained from real range image pairs show that the algorithm is robust and performs well in presence of occlusions and incorrect segmentations. Motion transformation between image frames is computed using the planar and the quadric surface pairings. A least-squares minimization procedure is formulated that estimates the best motion transform, subject to the constraints of rigid motion. For the case of linear feature pairings, the motion computation becomes tractable because the rotation and the translation computations become independent of each other. However, for quadric surfaces this is not true. The equation to be minimized is highly nonlinear and the uniqueness of solution cannot be guaranteed. The solution obtained computes the motion by extracting unique linear features from the quadric surfaces and using them to compute the motion transformation. The main contribution of the work is a surface-based framework for motion estimation from a sequence of range images. The primary issues of correspondence and motion computation are formulated and solved in terms of the surface descriptions.

Providing end-to-end QoS assurances in a CORBA-based system

In this paper we describe the implementation of our end-to-end QoS-driven resource management scheme, called ERDoS, within a CORBA-compliant ORB that we call the ERDoS QoS ORB. Unlike other real-time CORBA implementations that focus on real-time support for simple client-server applications, our ERDoS QoS ORB provides end-to-end QoS support (i.e., QoS spanning computer network, and storage resources) to applications, while retaining the benefits of an open distributed object system. Specifically we present three contributions. First, we present a model for describing end-to-end applications as a combination of client-server interactions between CORBA objects. Second, we define a model for relating the user/application level QoS requirements to the corresponding resource demand requirements of these individual CORBA objects. Third, we suggest a framework for performing distributed resource management within the CORBA environment.

Dynamic adaptation of video for transmission under resource constraints

A distributed system that supports applications with varying quality of service (QoS) requirements must use many adaptive mechanisms to guarantee performance to the applications. This paper presents the use of the mechanism of graceful adaptation for applications with multiple levels of performance. The different levels of performance use different amounts of resources. Our ERDoS architecture supports such mechanisms in order to provide end-to-end service guarantees to applications. The mechanism we present is based on market protocols in which trade-offs are made between the different QoS dimensions of an application so as to satisfy the applications QoS needs under resource constraints.

Adaptive QoS support for distributed, Java-based applications

Describes SRI Internationals ERDoS (End-to-end Resource management of Distributed Systems) project. ERDoS is a distributed middleware with comprehensive resource management support, including application scheduling, resource allocation and graceful adaptation. We discuss how we have implemented this resource management as part of the Java Virtual Machine (JVM). We have also provided a resource management application programming interface (API) as a new Java class library. The ERDoS development environment enables nonprogrammers to create complex distributed applications by simply stringing together multiple Java methods. The ERDoS runtime environment utilizes a structuring algorithm to create a distributed Java application dynamically, and then provides allocation, scheduling and content-based, graceful, adaptive QoS support to these applications. We demonstrate the benefits of content-based adaptation and of dynamic structuring of distributed applications, using a set of video applications.