Research On Intelligent Dynamic TDMA MAC Protocol For Ad Hoc Networks Based On Service Awareness

Wireless Ad Hoc Networks are widely used in military and civilian fields because of their distributed networking,high scalability and multi-hop transmission.Currently,the demand for communication services is increasing and the types of services are changing,and the traditional Medium Access Control（MAC）protocols cannot meet the differentiated needs of various services in many types of business scenarios.This thesis focuses on how to make full use of the limited wireless channel resources to meet the differentiated needs of multiple types of services,combining the characteristics of Ad hoc networks to design a distributed dynamic TDMA（Time Division Multiple Access）MAC protocol with differentiated service priorities for multiple types of service scenarios.Firstly,this thesis designs a service-aware priority preemption protocol PPTDMA（Priority preemptive TDMA）based on the Five-Phase Reservation Protocol（FPRP）,considering the impact of service priority on node access channels and reservation resources to meet the demand for differentiated network quality of service for multiple types of services,and realizing nodes to negotiate dynamically on demand to occupy time slot resources based on node priority.The protocol proposes improvements in two aspects.First,it proposes a priority preemption mechanism,in which nodes preempt the occupied information time slots of their neighbors according to the node priority and notify the reservation result to ensure the reservation advantage of high priority nodes.Second,the actual situation of node queues and multi-hop pseudo Bayesian algorithm are combined to design the node priority-based Random Access Algorithm（PRAA）,which distributes and dynamically adjusts the access probability of nodes with different priority levels to complete a conflict-free access channel within two hops.Simulation results show that the PPTDMA protocol can ensure that high-priority nodes have superior channel access and time slot resource occupancy,and the performance of high-priority services can also be prioritized when dealing with network burst services.Secondly,considering the balanced allocation of channel resources while guaranteeing the reservation demand of different priority nodes,this thesis proposes the time slot adaptive reservation algorithm DQN-TSARA（DQN-based Time Slot Adaptive Reservation Algorithm）combined with deep reinforcement learning DQN（Deep Q Network）algorithm.Based on this algorithm,the DQN-PPTDMA protocol is further designed.The number of time slots reserved by nodes is adaptively adjusted by considering the nodes’ own demand,the queue service capacity of neighboring nodes and the demand pressure of neighboring nodes’ time slots reservation,and the reservation result of information time slots is finally determined by combining with the priority preemption mechanism to achieve the purpose of balanced allocation of channel resources.Simulation results show that the DQN-PPTDMA protocol can make adaptive adjustments to the decision on the number of time slot reservations according to the changes in network load pressure,which ensures a balanced allocation of time slot resources while safeguarding the reservation advantage of high-priority nodes.Compared with the PPTDMA protocol there is an improvement in the balanced allocation of resources.