This dissertation addresses three important problems in wireless ad hoc and sensor networks. The subject of wireless ad hoc and sensor networks is identified as one of the most important areas of research for future wireless systems. The robustness, flexibility, and adaptability of these self-organizing networks unleash unprecedented opportunities for a wide spectrum of applications. In addition, wireless sensor networks provide information around our daily lives as well as information about the physical environment at a finer granularity, yet at a larger scale than has been possible before. Wireless ad hoc and sensor networks are changing the way we perceive and share information of all kinds. This thesis addresses three important problems in wireless ad hoc and sensor networks, i.e., (1) service differentiation, (2) the funneling problem, and (3) the mobile tracking problem. To solve these problems, this thesis presents three information-driven systems, which are MetroTrack, Funneling-MAC, and SWAN. The first contribution of this thesis is to propose SWAN, a stateless wireless ad hoc network model, which supports service differentiation for real-time and best-effort traffic in a simple, robust, and responsive manner. As a “stateless” model, the intermediate nodes in SWAN do not keep any per-flow or aggregate state information. Therefore, it is not necessary to establish, update, refresh, and remove per-flow state information. As a result, SWAN does not introduce any complex signaling and state control mechanisms that are required in “stateful” quality-of-service (QoS) approaches. Next, this thesis investigates the problem of funneling effect in wireless sensor networks. Wireless sensor networks exhibit a unique funneling effect that is a product of the distinctive many-to-one, hop-by-hop traffic pattern found in the networks. The funneling effect causes a significant increase in transit traffic intensity, collisions, congestion, packet losses, and energy drain. While congestion control and data aggregation techniques can help counter this problem, they cannot fully alleviate it. This thesis takes a different, but complementary, approach to solving this problem and presents the design, implementation, and evaluation of a localized, sink-oriented funneling-MAC capable of mitigating the funneling effect. The funneling-MAC is based on CSMA/CA being implemented network-wide, with a localized TDMA algorithm overlaid in the funneling region (i.e., within a small number of hops from the sink). The funneling-MAC represents a hybrid MAC approach that does not have the scalability problem associated with the network-wide deployment of TDMA. The funneling-MAC represents an information-driven approach in the sense that the sink node collects information about the traffic pattern and the intensity inside the funneling region and uses the information to dynamically compute the schedule and new depth of the funneling region. The final contribution of this thesis is to present a mobile tracking system using mobile phones carried by people as mobile sensors. This thesis introduces MetroTrack, a novel distributed tracking system that tracks mobile events using off-the-shelf mobile phones. MetroTrack presents the information-driven tasking algorithm to respond to dynamically moving targets and time-varying density of sensors around the target. The tasking algorithm is information-driven in the sense that each mobile sensor uses to make local decisions about whether to forward the task using local state information and sensor readings. MetroTrack also presents the prediction-based recovery algorithm that recovers the target when it is lost due to time-varying sensor coverage. MetroTrack recovers the lost target by tasking mobile sensors in close proximity to the lost target based on a prediction of its future location using a distributed Kalman-Consensus filter. A proof-of-concept prototype MetroTrack is implemented using Nokia N80 and N95 phones. The performance of MetroTrack, which cannot be fully analyzed in the small-scale, proof-of-concept testbed, is studied in large-scale simulations. The simulation results indicate that MetroTrack is robust in the presence of different mobility models and densities of mobile sensors. (Abstract shortened by UMI.)