Cloud Radio Access Network (C-RAN) has been proposed as a fifth generation (5G) cellular network that is designed to physically separate the Baseband Units (BBUs) from the Remote Radio Heads (RRHs). The BBUs are placed in the BBU pool/hotel, while the RRHs are located in the cells. This separation enables sharing baseband processing resources among RRHs by generating Virtual Machines (VMs) in the BBU pool, leading to radio coordination and energy saving opportunities. Nevertheless, the failure of a BBU can affect several RRHs, causing outages in the radio network. Therefore, designing a reliable and resilient C-RAN is essential to provide high availability and service continuity in case of BBU failure. This paper deals with using formal verification techniques to analyze the performance and the dependability of C-RAN architecture. We propose fault tolerant VM allocation strategies that can handle BBU failures, thus improving the system’s overall reliability. Based on redundancy techniques, these strategies ensure call recovery through migration to functional BBUs in the case of a BBU failure. In order to evaluate the mutual impact of performance measures on the system’s dependability, we develop performability (combination of performance and dependability) Markov Reward Models (MRMs) for the proposed strategies. We use Probabilistic Model Checking (PMC) to check the performability requirements written with Continuous Stochastic Reward Logic (CSRL). The proposed models are implemented and checked using the PRISM model checker, which supports solving MRMs and checking CSRL properties.

中文翻译：

---

云无线接入网容错虚拟机分配策略的形式化可靠性分析


云无线电接入网络 (C-RAN) 已被提议作为第五代 (5G) 蜂窝网络，旨在将基带单元 (BBU) 与远程无线电头 (RRH) 进行物理分离。 BBU 放置在 BBU 池/酒店中，而 RRH 则位于小区中。这种分离通过在 BBU 池中生成虚拟机 (VM) 来实现在 RRH 之间共享基带处理资源，从而实现无线电协调和节能机会。然而，一个 BBU 的故障可能会影响多个 RRH，从而导致无线网络中断。因此，设计可靠且有弹性的 C-RAN 对于在 BBU 发生故障时提供高可用性和服务连续性至关重要。本文涉及使用形式验证技术来分析 C-RAN 架构的性能和可靠性。我们提出了容错VM分配策略，可以处理BBU故障，从而提高系统的整体可靠性。这些策略基于冗余技术，在 BBU 发生故障时通过迁移到功能 BBU 来确保呼叫恢复。为了评估性能指标对系统可靠性的相互影响，我们为所提出的策略开发了可执行性（性能和可靠性的组合）马尔可夫奖励模型（MRM）。我们使用概率模型检查（PMC）来检查用连续随机奖励逻辑（CSRL）编写的性能要求。所提出的模型是使用 PRISM 模型检查器实现和检查的，该检查器支持求解 MRM 和检查 CSRL 属性。