Research On Power Diversity Of Random Access Schemes For IoT-Oriented Satellite Networks

With the rapid development of satellite communication and the emergence of the Internet of Things(Io T)on the ground,the satellite network which can support Io T applications has become a research hotspot.In the Io T-oriented satellite network,it can realize functions such as navigation,positioning,environmental monitoring and early warning of natural disasters in special areas by a large number of sensor nodes or machine type communication(MTC)terminals.Compared with the terrestrial Io T,the satellite Io T covers more areas,the system is harder to be destroyed and it promotes the intelligence of the information network.To avoid "signaling storms" and frequent resource scheduling processes,the employment of random access(RA)scheme is best for the access of the satellite network.However,it occurs severe packets collisions when massive Io T nodes try to access the network,so RA protocols with large throughput are urgently needed.By introducing the power dimension in the existing protocols to expand the channel competition space,the throughput of protocols can be improved.Different nodes select different power levels to send packets.The packet with larger power may still be detected by receiver due to the capture effect after packets collision.Therefore,more packets are successfully decoded.This thesis focuses on RA protocols with power diversity in Io T-oriented satellite network to improve the throughput,the main research contents are as follows: 1.The parameters affecting the throughput of the RA protocols with power diversity are optimized.First,we intriduce the contention resolution diversity slotted ALOHA(CRDSA)protocol with power diversity,where the power level selection probablity is optimized.Then for irregular repetition slotted ALOHA(IRSA)protocol with power diversity,the function between the throughput and the number of available power levels,the probability of power levels to be chosen by terminals are researched separately.By theoretical derivation and solution of optimization problem,the optimal parameters of IRSA protocol with power diversity are obtained.In order to solve the problem that the throughput of protocol under high traffic load is limited,CRDSA protocol with power diversity based on the adaptive load assisted scheme is researched.By deriving the relationship between the capture probability in one slot and the power level selection probability,a scheme is obtained for the terminal to adjust the power level selection probability in real time based on the current load information to maximize the capture effect.2.By introducing the reinforcement learning,a RA protocol with power diversity based on Q-learning is researched.First,we introduce the one-dimensional Q-learning RA protocol,which only exploits Q-learning on slots.Considering the delay characteristics of satellite Io T,a parity frame transmission scheme adapted to the large delay of satellite network is researched.Then we focus on the two-dimensional Q-learning RA protocol,which combines Q-learning not only in slots but also in transmit power levels.Each terminal adjusts the slot and power level simultaneously and autonomously according to the result of last transmission.After convergence,each device selects a dedicated slot to send packet at a dedicated transmit power level.Consequently,the packets collision is greatly avoided and the throughput of traditional protocols with power diverity will be improved.Aiming at solving the problem that the throughput decreases sharply after the load exceeds the saturation point,an access control scheme using Q-learning to adjust the access probability is researched.In this scheme,the access probability is adaptive based on whether the system is overloaded.The real-time load estimation with high complexity is no longer needed.The dynamic access control can make the RA protocol with power diversity based on Q-learning maintain high throughput when system is overloaded.