Abstract:In a cognitive radio network (CRN), the interference caused by secondary users (SUs) is conventionally regarded as an obstacle to the throughput of the primary network. However, when considering security, the interference brings potential benefits to the secrecy throughput of the primary network. For a stand-alone primary network, the secrecy guard zone, including the cooperative mode and non-cooperative mode, has been shown as an efficient method to enhance the network security. The cooperative mode is traditionally thought to outperform the non-cooperative mode due to the jamming effect of the artificial noise generated by primary transmitters on eavesdroppers. However, allowing SUs to access the licensed spectrum, the resulting interference can be used as a source of the artificial noise instead of that generated on purpose by primary transmitters, which in return benefits the energy efficiency of the primary network. Inspired by the two benefits brought by SUs, this paper considers a random underlay CRN with eavesdroppers overhearing primary transmissions. To enhance both the security and energy efficiency of the primary network, we apply a secrecy guard zone around each primary transmitter, and adopt the non-cooperative mode. We exploit the stochastic geometry to model such a random CRN and analyze the connection/secrecy probability of primary links. We propose a criterion for guaranteeing the performance of secure primary network. The main idea is that compared with the primary network with the cooperative mode, the access of SUs should not reduce the secrecy throughput and energy efficiency of the primary network with the non-cooperative mode. Based on the obtained analytical results, we design the optimal secondary link scheduling schemes under the criterion. Both analytical and numerical results show that the interference from SUs can be exploited toward a secure and energy-efficient primary network and provide SUs with extra transmission opportunities.

Interference-Limited Ultra-Reliable and Low-Latency Communications: Graph Neural Networks or Stochastic Geometry?

Graph Neural Networks for Quality of Service Improvement in Interference-Limited Ultra-Reliable and Low-Latency Communications

Design of Retransmission Mechanism for Decentralized Inference with Graph Neural Networks

A Graph Neural Network Based Decentralized Learning Scheme

Resource Allocation in URLLC with Online Learning for Mobile Users.

Message Passing Assisted Scalable Distributed Link Management for Ubiquitous Network

Optimal Wireless Resource Allocation with Random Edge Graph Neural Networks

Privacy-Preserving Decentralized Inference with Graph Neural Networks in Wireless Networks

Joint URLLC Traffic Scheduling and Resource Allocation for Semantic Communication Systems

Learning to Optimize Resource in Dynamic Wireless Environment Via Meta-Gating Graph Neural Network

Decentralized Learning Strategies for Estimation Error Minimization with Graph Neural Networks

Unsupervised Deep Learning For Ultra-Reliable And Low-Latency Communications

Deploying Graph Neural Networks in Wireless Networks: A Link Stability Viewpoint

A Neural Network Based Conflict-Graph Construction Approach for Ultra-Dense Networks

Edge-enhanced graph neural network for DU-CU placement and lightpath provision in X-Haul networks

Sparse Graph Neural Networks for Two-Timescale Wireless Resource Allocation.

Machine-Learning-Based Parallel Genetic Algorithms for Multi-Objective Optimization in Ultra-Reliable Low-Latency WSNs

Towards Secure and Energy-Efficient CRNs Via Embracing Interference: A Stochastic Geometry Approach.

Delay-Oriented Distributed Scheduling Using Graph Neural Networks

Joint Routing and Scheduling Optimization in Time-Sensitive Networks Using Graph-Convolutional-Network-Based Deep Reinforcement Learning.

Random Graph-Based Neuromorphic Learning with a Layer-Weaken Structure