Motivation : In recent years, researchers have developed a large and growing set of protocols and algorithms to improve the throughput and capacity of wireless networks. These schemes span the physical (PHY), medium access control (MAC), and higher layers of the protocol stack. Most effective innovations however require cross-layer modifications of both PHY and higher layers. To date, the verification of those designs has been limited to simulations or small setups relying often on off-line processing of the results. MAC layer results that rely on even the tiniest modification of the PHY can only be verified under simplified networking assumptions. Similarly, novel PHY algorithms are typically only verified for a single wireless link, avoiding complex scenarios. Most importantly, there is almost no cooperation between PHY and networking communities, as the tools and testbeds they use are incompatible. ; Contributions : In this paper, we propose a methodology for fully flexible PHY, MAC, and network layer verification that is designed to (a) reuse existing software components from PHY and network communities, (b) enable both simple- and expert-level modification and configuration of all components, (c) have real-time performance benchmarked with off-the-shelf systems, and (d) enable large networking experiments including off-the-shelf nodes for rapid experimentation, testing, and comparison. The main contribution of this paper is the introduction of an approach that enables the realization of full software-defined radio (SDR) sensor nodes, all running on a single field-programmable gate array and reusing PHY layer SDR tools and typical operating systems such as Contiki OS. Subsequently, the paper will illustrate the strengths of the proposed approach by demonstrating communication with off-the-shelf sensor nodes. This allows fair benchmarking with state-of-the-art or off-the-shelf solutions. Finally, some cross-layer improvements are proposed and compared with the baseline off-the-shelf system. This proves our claims that the proposed platform is a very useful tool for cross-layer experimentation, in that it allows full cross-layer control of the PHY and network layers, and moreover enables elegant comparison with state-of-the-art designs. This architecture is provided to the open source community (http://claws.be/), in order to become a framework for validating and benchmarking wireless cross-layer innovations.