Recent advances in wireless technologies, coupled with the concomitant abundance of portable devices, are opening up exciting opportunities for the future of wireless networking. In this context, MANETs (Mobile Ad hoc NETworks) can play an important role since they are autonomous, do no require any pre-existing infrastructure for communication purposes, can be conveniently deployed in locations where building a networking infrastructure is expensive or cumbersome, possibly to increase the reachability of infrastructure-based networks. In addition, ad hoc networks may operate whether in a standalone fashion, or be connected to other types of networks so as to provide users an access to a wide diversity of resources. However, MANETs are highly dynamic and suffer from frequent and unpredictable changes in the network topology due to the fact that devices are mobile and operate with low battery power. Considering these specific characteristics, it appears that existing approaches, mostly designed to operate in wi ed networks, cannot be adopted to locating and providing access to resources in pervasive computing environments. Indeed, these approaches are mostly centralized and are thus, unsuitable for infrastructure-less networks. This inefficiency, aggravated by the low connectivity and the mobility of nodes, remains a major obstacle in the development of applications for the wireless environment that integrates MANETs. Specifically, resource localization in pervasive computing environments has been an attractive area of research in wireless and/or mobile environments. However, proposed solutions are broadcast-based and result in overloading the constrained network. As a result, they do not solve the issue of scalability, mobility or adaptability. The issue addressed by this thesis is therefore to provide automatic and efficient resource localization and ubiquitous access to resources available in a pervasive computing environment integrating ad hoc network(s) with other interconnected network(s) (that can be infrastructure-based). This dissertation examines first the techniques and strategies that have been proposed in infrastructure-less environment. It then introduces a conceptual architecture to enable ubiquitous resource localization and access in infrastructure-less pervasive environments. This architecture is adapted to the characteristics of the ad hoc network (e.g., node density, network area). It leads us to tackle on separate way resource localization into small or scalable ad hoc networks. The first approach for resource localization is cooperative and fully distributed, and balances its performance against its cost by conveniently balancing communication load among the devices. This protocol includes some energy-aware caching and prefetching strategies that improve the overall performance of the resource localization. In order to support efficient resource localization in a scalable ad hoc network, we then introduce a semi-distributed solution that is based on a set of directories dynamically deployed over the ad hoc (or possibly hybrid) network. In order to improve the performance of the proposed solution, we introduce several cross-layer optimizations. The design of the resource discovery protocols is further complemented with extensive evaluation of performance, including both theoretical or/and practical studies based on simulation and implementation. Finally, we present the integration of our scalable resource discovery protocol in a middleware grounded on Web Service technologies. This middleware enables ubiquitous access to mobile Web services based on automatic discovery in dynamic pervasive computing environment.