| Power efficient routing is a challenging issue in sensor networks due to the limited power capacity of sensor nodes. In this dissertation, routing algorithms were explored to improve power consumption for common routing tasks by leveraging geographical information of sensor nodes without compromising the distributed and scalable routing constrain. The power efficient routing problem was divided into three scenarios: unicast, mulicast and broadcast, depending on the characteristics of the routing task.;In unicast routing, the radio range of transmissions has a significant effect on the power consumption of both the transmitting node and listeners. In response, a Geographical Power Efficient Routing (GPER) protocol, which only uses local information, is presented. A set of probabilistic multipath routing algorithms, which can generate braided multipaths, was investigated to slow the power drain on common paths.;In multicast routing, the focus was on the challenge of demand-scalable multicast routing in wireless sensor networks. To address this, a fully distributed and stateless Geographic Multicast routing Protocol (GMP) for wireless sensor networks is presented. The multicast protocol was further extended to cases where the data transmission could scale based on the demand. This led to three new weighted multicast routing algorithms: the weighted Geographic Multicast Protocol (wGMP), the weighted Location Guided Steiner tree (wLGS) and the weighted Position Based Multicast (wPBM).;In the broadcast scenario, problems in existing flooding based and pruning based broadcast algorithms were studied and as a result, a Geographic Power Efficient Broadcast algorithm was introduced to reduce the overlaps of broadcast coverage area and thus, total power consumption. This algorithm is a hybrid protocol that combines broadcasts of maximum transmission radio range and forwards of much smaller transmission radio range. Through extensive simulation, the performance of the proposed routing algorithms were studied and showed that significant power consumption improvement could be obtained compared to other pre-existing algorithms. |
