| With the global deployment and standardization of the fifth-generation(5G)broadband wireless network from 2020,human wireless communication activities have taken another step towards the era of massive device connection.However,the existing technologies in 5G are not enough to meet the requirements of network deployment for the next decade.In order to solve the limitations of 5G networks as well as realize the era of the Internet of Everything with higher speed and wider system coverage to meet the needs of intelligent mobile communications in 2030 and beyond,the beyond five generation and sixth-generation(6G)wireless mobile network systems will have new paradigm shift in communication services.The service vision of the next-generation network system mainly includes the following four aspects.1)The air-space-sea-ground global network coverage can be realized by integrating satellite communication,drone communication,and maritime communication.2)All spectrum resource will be explored to provide larger data capacity,including radio frequency wireless communication frequency bands and optical wireless communication frequency bands.3)Artificial intelligence and machine learning technology are adopted to realize the communication service requirements of automated management and intelligent application scenarios.4)The security of the network system should get a strong guarantee,when introducing the industrial vision of the Internet of Everything.To meet the network requirement for 6G,new air interface transmission technologies and novel network architectures should be developed to realize aforementioned new paradigm shifts in 6G,namely as global coverage,all spectrum,full applications,and strong security.This paper boils down the full coverage network system as a cooperative communication system.Based on this model,this work focuses on investigating the application of non-orthogonal multiple access technology(NOMA)in the relay cooperative network,and proposes a hybrid visible light/radio frequency(VLC/RF)communication network with NOMA user cooperation to explore the full spectrum resource,along with exploring the cognitive coordinated system in VLC/RF network with considering the concept of simultaneous lightwave information and power transfer.Moreover,a fully-decentralized federated learning framework is proposed to realize the data privacy protection of multi-point cooperative communication network.Firstly,in the cooperative NOMA system with dedicated-relay,where the signals are transmitted with the aid of either decode-and-forward(DF)or amplify-and-forward(AF)relay,exact expressions for outage probability and ergodic capacity are derived.Based on the above analysis,this work proposes an adapative forwarding strategy selection scheme for both fixed user allocation and optimized power allocation.The presented simulations verify the accuracy of the analysis,and the efficiency of the proposed forwarding strategy.In addition,based on the two-hop model of DF-NOMA,a half-duplex two-hop scheme with unequal time slots is proposed.By optimizing time slot resource allocation and user power allocation at the same time,a better user performance can be obtained than the traditional equal time-slot system.Secondly,an indoor lightwave downlink wireless communication network with the NOMA technology is proposed,which consists of one visible light communication-access point(VLCAP)and a pair of randomly located users.Although both users can directly receive information from the VLC-AP,the performance of the far user is degraded compared to the near one due to the asymmetrical channel gains.Thus,the user cooperation strategy is proposed to improve the performance of the far RF user by using mixed VLC/RF relaying technique in parallel with the wireless optical direct link.To efficiently exploit the available heterogeneous links,the concept of cross-band selection combining is introduced at the receiver side,according to which the far user is continuously served by either the mixed VLC/RF or the direct VLC link.Furthermore,the performance of the proposed scheme is thoroughly investigated and compared to appropriate baselines.To this end,we derive closed-form expressions for the outage probability of each user as well as the system sum throughput.Finally,simulation results are provided to verify the effectiveness of the proposed scheme and the accuracy of the corresponding analysis.Then,the concept of cognitive radio and the simultaneous lightwave information and power transfer(SLIPT)are introduced into the hybrid VLC/RF cooperation system,where the VLC user transmit information to RF users through a cooperative forwarding strategy,while considering that the VLC user adopts the proposed SLIPT strategy.In addition,the system uses a cognitive-based strategy for adaptive resource allocation,and then accesses the radio frequency user’s information reception without affecting the performance of the VLC user.Furthermore,by taking into account the random deployment of the VLC and RF user terminals and a tractable approximation for the harvested energy,the closed-form outage probability for the VLC user is given,while for RF user,the corresponding expressions are approximately presented in both infinite-series and closed-form.Moreover,under the constraints based on the cognitive resource allocation strategy,the outage probability of RF user can be further minimized by optimizing the direct current/alternative current(DC/AC)component,which can be specifically realized by tuning the intensity modulation-direct detection scheme at the transceiver structure.Finally,the numerical results verify the compactness of the upper and lower bounds of the proposed energy harvesting,the accuracy of the outage probability results,and the effectiveness of the proposed DC/AC parameter configuration.Finally,a decentralized network framework is proposed to solve issues in the traditional communication model,and the federated learning model is built from the multi-user node cooperative communication,in order to reduce the burden on the central server while maintaining information security when large-scale user data is connected to the network.Moreover,the utilized machine learning framework can help to find robust communication solutions,which can avoid the impact of environmental factors on communication quality to a certain extent.Combining the above technologies,a random parallel random walk alternating multiplier direction algorithm is proposed to solve the task of coping with decentralized federated learning.This method can be applied while maintaining high communication/learning efficiency,and achieve enhanced privacy protection.Furthermore,the framework can also cope with the challenges of time-varying connected graphs and random data collection,which has the potential for rapid convergence,so it is very suitable for dynamically connected network topology nodes.Next,the specific communication scenario is utilized to verify the practicability of the proposed algorithm framework,that is,considering the dynamic UAV swarm to complete the robust design of beamforming,at the aid of the extreme learning machine model.Eventually,through the iterative update of collaborative information among multi-user nodes,all users can adopt the same machine model to achieve a global consensus solution. |
