Context-Aware Resource Management for End-to-End QoS Provision in Service Oriented Applications*

Shudong Chen1, Johan Lukkien1, Richard Verhoeven1, Pim Vullers1, and Goran Petrovic2
1 Department of Mathematics and Computer Science,
Eindhoven University of Technology, The Netherlands.
2 Department of Electrical Engineering,
Eindhoven University of Technology, The Netherlands.
Electronic devices admit the composition of applications from distributed services, like shared content, functionality and resources. Without proper management the quality of service (QoS) of this type of application is unacceptable. This paper presents a context-aware resource management middleware for service oriented applications, which aims to handle the inherent dynamics of services and the network. A way to manage the services and resources of a device is presented. Based on measurable and manageable services, this middleware can cope with changes of the environment of an application and optimize its overall performance. User desired application-oriented QoS is specified by an Orchestrator, which orchestrates the collaboration of distributed services to accomplish an application. A resource management service is built to support making dynamic service coordination decisions for the Orchestrator. For this, the resource management service gathers statistics of the required services and their resource utilization over a cross-layer monitoring model. An important aspect for keeping the privacy of the services and controlling the access to resources, all devices and services are organized in virtual communities. Only authenticated users are capable to access services and resources. A free view point 3D video streaming application is chosen to prove the feasibility of the proposed middleware.
access control, resource management, service-oriented applications
* This work is supported by the research project of Freeband I-Share: Intelligent Middleware for Sharing Resources for Storage, Communication and Processing of Multimedia Data, supported by the Dutch government.

1   Introduction

There is a rapidly increasing amount of resource constrained devices, e.g. PDAs, CE equipments that expose their capabilities as networked services leading to the emergence of ubiquitous computing applications, like content, functionalities and resources sharing. The applications are achieved through the composition of services. Without proper management this composition has limited quality and does not scale beyond limited numbers of application. For example, in a multimedia streaming application over an in-home network, the video can be smoothly streamed to a PC with a 100Mbps Internet connection. However, when the streaming is switched to a PDA, the video may suffer from the error-sensitive wireless connection and the constrained resources of the PDA. Ubiquitous computing applications that depend on available services need to cope with issues like heterogeneity of the networks, mobility and capabilities of devices etc. It becomes even more challenging when we consider co-existence of different traffic, for instance, a video provider that has to respond to multiple invocations simultaneously. Therefore, dynamic resource management is crucial to support the end-to-end QoS of an application.

This paper presents a context-aware resource management middleware designed for service oriented applications, supporting dynamic application QoS adaptation. This presented middleware was developed within the context of the I-Share project [1] that focuses on the secure and efficient sharing of multimedia through forming applications from distributed services in virtual communities. More specifically, the proposed middleware aims at:

  1. Providing a secure service discovery, access and collaboration environment;
  2. Assisting end users to form applications through external service orchestration;
  3. Supporting application-oriented end-to-end QoS provision through performance monitoring and prediction of both services and the network, and performing QoS adaptations when the environment of the application alters.

Based on trust between service providers and service users, services provided by devices are organized into virtual communities (VCs). Only authenticated users are capable to access VC services and resources. Service discovery, access, and collaboration all happen in the scope of a virtual community. Services can be bound by an Orchestrator at runtime to accomplish applications. A resource management service supports making adaptive service coordination decisions for the Orchestrator. For this, the resource management service predicts the performance of the services through a cross-layer monitoring model which gathers real time resource utilization of these services.

We compare our work with several cross-layer resource management platforms [2, 3]. The largest concern raised in those platforms is their scalability when expanding to different application domains or networks. Work in [2] focuses on end-to-end QoS provision in multimedia content delivery and it requires the media content to support an MPEG-21-based codec. Our work supports QoS assurance for a great variety of applications. The system proposed by [3] assumes that a centralized manager negotiates monitoring operations with each domain along the service delivery path. This results in a scalability problem when the inter-domain network expands. Our middleware provides a solution for measuring the QoS which eliminates the inter-domain issues.

The paper is organized as follows: Section 2 presents the virtual community based service discovery and access control. The design of the context-aware resource management is detailed in Section 3. Section 4 describes a 3D video streaming scenario to validate the feasibility of the proposed middleware. Finally, conclusions are drawn in Section 5.

5   Conclusions

This paper presents VICSDA, a context-aware dynamic resource management middleware for service oriented applications, which aims to handle the inherent dynamics of the local devices and the network. Based on a secure service cooperation environment provided by forming services into virtual community, a cross-layer monitoring architecture is designed to gather the performance statistics of services and the network. The ResourceMgt service can interpret applicationoriented QoS requirements into performance-oriented metrics. Being aware of the changes in the environment of an application through an event publish/subscribe scheme, it can anticipate the performance of required services and the network by making service coordination decisions. Service performance prediction is executed at a soft state level to avoid heavy kernel load with respect to get a high precision. With the full control of VC services and their underlying resources achieved by a DevMan service, we can optimize the overall performance of a device and in turn of the application. The resource management service is platform independent. Benefit from the context-free representation method adopted for characterizing various parameters, the resource management service is flexible over a great variety of applications.

A 3D video streaming application is implemented to test the feasibility of VICSDA. Concrete outcome shows that this context-aware dynamic resource management middleware optimized the displayed video quality over adapting the 3D video's transcoding and delivery protocol based on the capacity of the delivery channel and of the display devices. Conclusions can be drawn that VICSDA supports end-to-end QoS provision to extensive service cooperation while keeping the authority, confidentiality, and full control of each service provider over its underlying resources.

Fault detection and recovery of the system will be addressed in the future to facilitate enhanced reliability and robustness of VICSDA.

References

  1. The Netherlands Freeband I-Share research project: http://www.win.tue.nl/san/amosa/ishare/intro
  2. A. Mehaoua, T. Ahmed, et al, "Service-driven Inter-domain QoS Monitoring System for Large-scale IP and DVB Networks", in ELSEVIER Computer Communication Journal, Special Issue on Monitoring and Measurements of IP Networks, June 2006, Volume 29, Number 10, page 1687-1695.
  3. European IST-INTERMON projects: http://www.ist-intermon.org/
  4. S. Chen, J.J. Lukkien, I. Radovanovic. Freeband I-Share research project deliverable 1.5. March 2007.
  5. P.Th. Eugster, P.A. Felber, R. Guerraoui and A.-m. Kermarrec. "The Many Faces of Publish/Subscribe", ACM Computing Surveys, Vol. 35, No. 2, June 2003, page 114-131.
  6. S. Chen and J.J. Lukkien, "Secure Resource Control in Service Oriented Applications", Accepted by the 6th annual IEEE Consumer Communications & Networking Conference (CCNC 2009), in process.
  7. K. Lai and M. Baker, "Measuring Link Bandwidths Using a Deterministic Model of Packet Delay", ACM SIGCOMM Computer Communication Review, Volume 30, Issue 4, October 2000, page 283-294.
  8. S. Chen, J.J. Lukkien, P.H.F.M. Verhoeven, R.P. Bosman, M. Tjiong, Freeband I-Share research project deliverable 1.16. December 2007.
  9. J.A. Stankovic, M. Spuri, K. Ramamritham, G.C. Buttazzo, "Deadline Scheduling for Real-Time Systems: EDF and Related Algorithms", The Springer International Series in Engineering and Computer Science, Vol. 460, 1998.
  10. G. Petrovic, and P.H.N. de With, "Near-future Streaming Framework for 3D-TV Applications", In the Proc. of the IEEE International Conference on Multimedia & Expo (ICME) 2006, page 1881-1884.