Enabling Time-Synchronized and Interference-Aware Initialization of Wireless Sensor Networks

Abstract

Wireless sensor networks (WSNs) provide ad hoc wireless infrastructure to spatially distributed sensors to interact with physical or environmental phenomena. WSNs can offer a multitude of applications given that the sensors are able to collaborate and self-organize. These requirements are essential to accurately capture and reliably fuse the observations towards the application logic. This thesis studies time synchronization and interference management schemes to enable collaboration and self-organization in low-power WSNs.

Network-wide time synchronization is required both for the concurrent actuation of sensors and reliable data aggregation. Time synchronization is achieved by a clock synchronization algorithm which estimates the clock offset and clock skew at a sensor with respect to a reference time. The reference time is diffused in the network by a messaging protocol. This thesis studies clock offset and skew estimation methods for broadcast-based exchange of the reference time. The offset estimation is based on a study to eliminate the delay factors in the communication path. For skew estimation based on linear-regression, the correlation between time synchronization period and regression size is studied. In addition, a maximum-likelihood skew estimator, which minimizes the estimation error variance, is validated.

From an application's perspective, the time synchronization service should provide the desired synchronization accuracy in a transparent and energy-efficient manner. This thesis demonstrates this ability by extending the proposed time synchronization methods for a) tight synchronization among vibration samples in a structural health monitoring application, b) communication scheduling in time and frequency.

WSNs deployed in shared unlicensed bands need to analyze and mitigate interference. The low-power transmissions of sensor nodes are otherwise prone to corruption from high-power transmissions of coexisting wireless networks. Therein, this thesis proposes coexistence models for energy-detection based link-quality estimation. The coexistence models are utilized to formulate low-complexity coexistence enhancement algorithms named channel ranking. A ranking algorithm creates an ordered list of the candidate channels using a channel quality metric (CQM). The algorithms differ with respect to their design of CQM, the main design factor being the availability of network connectivity information which is usually unknown upon network initialization.