This thesis introduces techniques that exploit predictable mobility of the observer (data sink) to reduce communication range, thereby saving power in wireless sensor networks (WSNs). The movement of public transport vehicles is predictable, and such vehicles can act as mobile observers in wide area WSNs for applications such as pollution monitoring. To understand the gains due to mobility in single hop WSNs, data collection is modelled as a queuing process, with random arrivals representing randomness in the spatial distribution of sensors. This queuing model aids in analyzing the success in data collection and in quantifying network power consumption. A communication protocol is proposed which achieves the predicted power savings. In multihop WSNs, mobility reduces the expected number of hops from sensors to the observer, leading to significant power savings. In addition, having dedicated high data-rate relays near the path of the observer reduces power by orders of magnitude.