I-AdMiN: a framework for deriving adaptive service configuration in wireless smart sensor networks

Facilitating application development for distributed systems has been the focus of much research. Composing an application from existing components can simplify software development and has been adopted in a number of domains such as wireless sensor networks, mobile computing, ubiquitous systems, cloud computing, etc. E fficient application development in wireless smart sensor networks (WSSNs) generally faces more restrictions and is the focus of this thesis. Inherent limitations of wireless sensor networks such as memory size, bandwidth, computational capacity, and energy have driven WSSN application development towards low-level programming approaches which provide e fficiency but hinder sharing and reuse. Varying environmental conditions, faults, and changing application requirements are also common in long-term deployments of WSSNs. Environmental conditions and faults are important considerations in this domain since they can affect the availability of resources such as energy. For example, a stretch of cloudy weather can affect the energy availability of sensor nodes that are equipped with solar panels. On the other hand, requirements of WSSN applications vary considerably and can include energy consumption, time synchronization error, packet loss, etc. The increased dynamicity and complexity of WSSN applications require open systems that interact with their environment while addressing application constraints and hardware limitations. Our goal is to facilitate WSSN application development by allowing component sharing and reuse and dynamicity. Due to the importance of energy management on the lifespan of WSSN applications, our primary focus is on optimizing energy consumption while satisfying constraints that are derived from application requirements. We model applications as a composition of services. Services are self-contained software components with self-describing interfaces that represent their inputs and outputs as well as their non-functional properties. We illustrate the need for service sharing and dynamic service composition and their challenges through examples of real-world applications, namely structural health monitoring (SHM) and environmental and agricultural monitoring. In fact, our experience in the design, development and implementation of these applications that resulted in our eff ort to build a framework that facilitates software development for WSSN applications. We have developed middleware services that are deployed in two main testbeds. On the rst testbed, the Jindo Bridge in Korea, 113 nodes are deployed for long-term monitoring of structural health. The second testbed aims at environmental observation (soil moisture and nitrate) in a 40 acre fi eld in Champaign, Illinois that has 4 types of vegetation. The proposed solutions can be divided into three parts. First, we design a framework called I-AdMiN, which provides component deployment to enable dynamic service composition and adaptive recon figuration, while respecting the resource constraints and efficiency requirements of wireless sensor networks. Second, we address the eff ect of deployment characteristics and environmental conditions by dynamically deriving energy characteristics of services that comprise the WSSN application. This is done in a component called Monitor by using aggregate information on system energy consumption. Dynamic and on-line pro ling of services is important for two main reasons: i) many service characteristics such as energy consumption cannot be accurately determined until the full-scale deployment of the service, and ii) dependency relationships between diff erent services and between the hardware platform and services can aff ect the overall behavior of the system and must be taken into account in the course of service selection. Many such dependencies cannot be determined apriori and depend on the environment and run time characteristics. Finally, we design and implement a system called S4 to enable automatic selection of components and parameters to satisfy application requirements. S4 derives a constraint satisfaction problem from application constraints and service specifi cations and solves it to derive a selection of available services that form the application. Whenever available, S4 leverages dynamic information from the Monitor on service energy characteristics to optimize the energy consumption of the sensor network.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2014

Author dc:creator

• Moinzadeh, Parya

Contributors dc:contributor

• Agha, Gul A.
• Abdelzaher, Tarek F.
• Vaidya, Nitin H.
• Chandra, Ranveer

• Wireless Smart Sensor Networks
• Middleware Framework
• Energy Consumption
• Service Composition

Statement dc:rights

• Copyright 2013 Parya Moinzadeh

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/46767

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/46767