The paper introduces the concept of directed diffusion for wireless sensor networks, which is a data-centric communication paradigm where all interactions are for named data. In this paradigm, all nodes are application-aware and can cache and process data in-network to achieve energy savings. The authors explore and evaluate directed diffusion for a simple remote-surveillance sensor network, demonstrating that it can achieve significant energy savings and outperform traditional schemes like omniscient multicast under various scenarios. The evaluation includes both analytical and experimental results, showing that directed diffusion can robustly deliver data, adapt to empirically good paths, and aggregate responses to queries, leading to improved performance and energy efficiency. The paper also discusses the design choices and implementation details of directed diffusion, highlighting its key features and potential applications in sensor networks.The paper introduces the concept of directed diffusion for wireless sensor networks, which is a data-centric communication paradigm where all interactions are for named data. In this paradigm, all nodes are application-aware and can cache and process data in-network to achieve energy savings. The authors explore and evaluate directed diffusion for a simple remote-surveillance sensor network, demonstrating that it can achieve significant energy savings and outperform traditional schemes like omniscient multicast under various scenarios. The evaluation includes both analytical and experimental results, showing that directed diffusion can robustly deliver data, adapt to empirically good paths, and aggregate responses to queries, leading to improved performance and energy efficiency. The paper also discusses the design choices and implementation details of directed diffusion, highlighting its key features and potential applications in sensor networks.