Universal data access for run-time resource management in resource constrained wireless networks

Doctoral thesis English OPEN
Rerkrai, Krisakorn (2012)
  • Publisher: Publikationsserver der RWTH Aachen University
  • Subject: Kommunikation | Funknetz | Drahtloses Sensorsystem | Rekonfiguration | Prototyping | Leistungsbewertung | Energieeffizienz | Ingenieurwissenschaften | drahtlose Netzwerke | Radioressourcenverwaltung | generische Schnittstellen | nutzenbasierte Optimierungen | wireless networks | radio resource management | generic interfaces | utility-based optimization
    • ddc: ddc:620

Resource-constrainedwireless networks, e.g.wireless sensor networks (WSNs), small embedded devices with limited computational power and energy, have been the subject of intense research in the past decade. Moreover, recent technological advances and growing demand for better efficiency have led to a great number of link and network protocols for WSNs. The protocols depend on specific interfaces to exchange necessary information. Unfortunately these interfaces are often proprietary and highly different, thus making it harder to port different protocols and services from one platform to another. In fact, these differences are sometimes even making protocol interoperability difficult to achieve. The underlying heterogeneity clearly has limited interoperability and portability. Generic interfaces have the potential to enable enhanced portability and compatibility of different implementations. By providing powerful data accessing capabilities, these interfaces enable applications to monitor the system and react to the changes in the communications medium or environment and thus adapt on time accordingly. Seamless communication between software entities at different layers and different radio technologies is also a core functionality of future radio resource management (RRM) in the case of software defined radios. In this thesis we present the generic application programming interfaces (APIs), namely, lightweight version of the original unified link-layer API (ULLA), common application requirement interface (CAPRI), and universal data access engine (UDAE). We implement these frameworks on real hardware platforms and demonstrate their feasibility through several experimental WSN and WLAN testbeds. These generic frameworks can be used not only by applications running locally on nodes but also remotely by applications running in end-user devices. Firstly, the ULLA offers a common interface to retrieve link layer information independently of the deployed radio technology considerably simplifying development process of link-aware protocols and applications. The ULLA also offers a powerful notification mechanism, which can be used to monitor the network and trigger certain actions when the communication environment changes. Secondly, we have extended the concept of link-layer abstraction to generic data abstraction framework, UDAE. It is fully extensible, allowing data sources and users to be added flexibly at run-time. UDAE is implemented on two operating systems (Contiki and TinyOS).While the TinyOS implementation has been done as native code, the Contiki implementation is based on a reconfigurable component-oriented middleware enabling developers to fully benefit from the flexibility offered by the UDAE. Thirdly, CAPRI allows applications to register their utility functions to a radio resource controller at runtime and in automatic way. This enables the implementation of utility-based optimization and run-time reconfiguration of applications. In contrast to traditional utility-based optimization approaches that usually aim at optimizing the performance of a single application or service, our approach aims to generalize and express the requirements of multiple applications. Fourthly, we have applied the ULLA framework also to wireless sensor networks as a data acquisition mechanism. The sensor ULLA allows very versatile data collection and provides also capabilities for sensor networks to reorganize in the case of changes in the communications lines. Finally, we show how we can exploit functionalities of the generic interfaces in order to develop an optimization framework that allows the implementation of utility-based and cross-layer optimization for resource-constrained wireless networks. All the reference implementations have been made publicly available under an open source license.
Share - Bookmark