Performance Analysis of MIMO CR Multihop Relaying with Imperfect CSI in Short-Packet URLLCs

This paper evaluates the multiple-input multiple-output underlay cognitive multihop relay networks with short-packet communications, where general and practical scenarios are considered with multiple primary users and imperfect channel state information of the interference channels. For performance evaluation, the closed-form expressions of the end-to-end (E2E) block error rate for the considered systems are derived under consideration of quasi-static Rayleigh fading channels and the finite-blocklength (FBL) regime, from which the E2E throughput, energy efficiency (EE), latency, and reliability are also studied. Based on the analytical results, we adopt a machine learning-aided estimator, i.e., extreme gradient boosting (XGB), to predict the E2E throughput, EE, latency, and reliability for real-time configurations. The XGB-based evaluation achieves equivalent performance while significantly reducing the execution time compared to conventional analytical and simulation methods, which makes XGB an efficient tool to estimate the system performance in real-time applications.