Open-Radio Access Network (O-RAN) has become an important paradigm for 5G and beyond radio access networks. This paper presents an xApp calledxSlicefor the Near-Real-Time (Near-RT) RAN Intelligent Controller (RIC) of 5G O-RANs.xSliceis an online learning algorithm that adaptively adjusts MAC-layer resource allocation in response to dynamic network states, including time-varying wireless channel conditions, user mobility, traffic fluctuations, and changes in user demand. To address these network dynamics, we first formulate the Quality-of-Service (QoS) optimization problem as a regret minimization problem by quantifying the QoS demands of all traffic sessions through weighting their throughput, latency, and reliability. We then develop a deep reinforcement learning (DRL) framework that utilizes an actor-critic model to combine the advantages of both value-based and policy-based updating methods. A graph convolutional network (GCN) is incorporated as a component of the DRL framework for graph embedding of RAN data, enablingxSliceto handle a dynamic number of traffic sessions. We have implementedxSliceon an O-RAN testbed with 10 smartphones and conducted extensive experiments to evaluate its performance in realistic scenarios. Experimental results show thatxSlicecan reduce performance regret by 67% compared to the state-of-the-art solutions. Source code is available athttps://github.com/xslice-5G/code

Low Probability of Intercept (LPI) wireless communication is a critical aspect of modern wireless technology. Despite various techniques developed for LPI wireless communication, most of them require some form of pre-existing knowledge (e.g., encryption keys) or specific information (e.g., eavesdropper location and channel details). In this paper, we propose a novel physical-layer precoding technique called spatial pilot perturbation (SPP) to achieve efficient LPI wireless communications. Unlike existing methods, SPP operates without the need for pre-shared information between the two communication devices, nor any knowledge about potential eavesdroppers. It remains transparent to users and thus backward-compatible with off-the-shelf 5G/WiFi user devices. The core idea of SPP is to use different precoders for pilot and data symbols in a signal frame at the physical layer. Through a systematic precoder design, the pilot and data symbols will experience identical compound channels upon arrival at intended receivers, but experience different compound channels when intercepted by eavesdroppers. Consequently, the intended receivers can demodulate the signal frame, while eavesdroppers cannot. We have implemented SPP on 5G and WiFi testbeds and evaluated its performance through over-the-air experiments. Extensive experimental results show that SPP achieves an eavesdropping rate of ≤0.2% for 5G and ≤0.9% for WiFi, both at the cost of less than 18% throughput.