Abstract
As did it Internet and wireless communications some decades ago, the advent of sensor networks prefaces a revolution in our way of life. But before seeing these networks reach a democratization degree identical for example to the mobile phones one, some problems must be solved. To traditional constraints of ad hoc networks, should be added hardware limitations of sensors such as computing power, memory size and especially energy consumption, which make existing algorithms unsuitable. This thesis addresses two problems : localization and routing. On the localization side, a family of three methods is proposed to estimate sensor positions by associating error bounds, out of some sensors known exact localizations and depending to their measure capacity. Then, this family is extended to mobile sensor networks. The routing problem occurs when a sensor detects an event and wants to alert a base station. Routing strategies known as geographical are based on exact sensors positions. However, in practice, these positions are seldom accurate. This thesis proposes respectively two routing algorithms intended for static or mobile sensor networks by considering estimated positions, making these methods compatible with the localization algorithms.