Adaptive duty cycling in mobile sensor networks.
Doctoral thesis, UCL (University College London).
Mobile wireless sensor networks have recently attracted considerable attention. In particular, there is a significant interest in applying mobile sensor networks to wildlife and environmental monitoring, medical, and human-centric applications. Energy is a critical factor for most real deployments of sensor networks: many mobile applications, such as environmental monitoring ones, require months of unattended operation of large number of small battery-operated nodes. Due to slow advancements in battery and energy harvesting technologies, energy efficiency will remain an important issue for a long time. The general approach to energy saving in wireless sensor networks is to coordinate the wake-up time of nodes to maximize their sleep time while achieving application goals such as low latency or high throughput. A number of duty-cycling solutions have been proposed for static sensor networks. These solutions often assume a fixed topology and use scheduling techniques to coordinate the wakeup of nodes depending on traffic flows. However, in some applications, a fixed network topology cannot be assumed as some sensors are mobile: duty cycling in mobile networks is challenging because nodes need to continuously scan for neighbours, which is an energy intensive process. This thesis investigates the issues related to duty cycling of mobile wireless networks. We argue that duty cycling in mobile networks has to be adaptive to both mobility and traffic patterns. The thesis presents a two-level approach which exploits temporal connectivity patterns and offers practical techniques for duty cycling in sparse and dense scenarios. At macro level, the uncertainty of node discovery is a primary reason of power consumption, which causes the mobile nodes to periodically scan or listen for neighbours, draining battery. At this level, the approach is based on adapting the node discovery procedure to temporal activity patterns inherent to human-centric and animal-centric applications. At micro-level, the uncertainty of packet arrival is a major source of power consumption, and the approach mitigates this by using short-term synchronization, which constrains the packet arrival time to predefined time slots. The evaluation of the approach is performed through both simulation & implementation and deployment on a real sensor testbed. In particular, the performance of the macro level protocol is evaluated through simulation with real human and animal mobility traces and through deployment in Wytham Woods (Oxford) for badger tracking. The performance of micro-level protocol is evaluated through measurements on a small scale testbed.
|Title:||Adaptive duty cycling in mobile sensor networks|
Archive Staff Only