Software Defined Radio (SDR)-enabled cognitive radio network architectures are expected to play an important role in the future 5G networks. Despite the increased research interest, the current implementations are of small-scale and provide limited functionality. In this paper, we contribute towards the alleviation of the limitations in SDR deployments by developing and evaluating a resource allocation approach for cognitive radios implemented with SDR technology over two testbeds of the ORCA federation. Resource allocation is based on a Markov Random Field (MRF) framework realizing a distributed cross-layer computation for the secondary nodes of the cognitive radio network. The proposed framework implementation consists of self-contained modules developed in GNU Radio realizing cognitive functionalities, such as spectrum sensing, collision detection, etc. We demonstrate the feasibility of the MRF based resource allocation approach and provide extensive results and performance analysis that highlight its key features. The latter provide useful insights about the advantages of our framework, while allowing to pinpoint current technological barriers of broader interest.

中文翻译：

---

一种用于认知无线电网络的软件定义无线电跨层资源分配方法：从理论到实践

支持软件定义无线电 (SDR) 的认知无线电网络架构有望在未来的 5G 网络中发挥重要作用。尽管研究兴趣增加，但当前的实现规模较小且功能有限。在本文中，我们通过开发和评估在 ORCA 联盟的两个测试平台上使用 SDR 技术实施的认知无线电的资源分配方法，为缓解 SDR 部署中的限制做出了贡献。资源分配基于马尔可夫随机场 (MRF) 框架，为认知无线电网络的辅助节点实现分布式跨层计算。提议的框架实现由在 GNU Radio 中开发的独立模块组成，用于实现认知功能，例如频谱感知、碰撞检测等。我们证明了基于 MRF 的资源分配方法的可行性，并提供了突出其关键特征的广泛结果和性能分析。后者提供了有关我们框架优势的有用见解，同时允许查明当前具有更广泛兴趣的技术障碍。