Research On Semi-Distributed Load Balancing Routing Algorithm Based On LEO Satellite Network

Under the impetus of the global information revolution,the Low Earth Orbit(LEO)satellite network has the advantages of global coverage,not being affected by terrain and not being affected by land disasters,which can be used as a necessary extension and supplement to the traditional ground network.Due to the differential distribution of global traffic and the high dynamic characteristics of the satellite network topology,the load in LEO satellite network is unbalanced.Therefore,it is particularly critical to design a reasonable load balancing routing algorithm to improve the transmission capacity of the whole network.There are two obvious defects in the existing routing algorithms of load balancing in LEO satellite networks.Firstly,in the aspect of load balancing strategy,the existing strategies do not make full use of link information and have poor response ability to congestion,resulting in inefficient and unreliable routing strategies.Secondly,from the perspective of routing function implementation,the centralized global load balancing routing algorithm has high load perception overhead and slow response to congestion.Distributed local load balancing has poor shunt capacity and is easy to fall into local optimum.Therefore,it is very important to design a load-balancing routing algorithm which is suitable for LEO satellite network with high efficiency,reliability,strong shunt ability and timeliness.Aiming at the shortcomings of existing routing algorithms for load balancing in LEO satellite network in terms of strategy design and Routing function realization,this paper proposes Priority Forwarding Policy Based on Congestion Notification and Link State(CLPFP)and Semi-Distributed Load Balancing Routing algorithm SDLBR),and the CLPFP and SDLBR were modeled and simulated using OPNET software,and the performance was compared with other related algorithms.The main research content is divided into the following three parts:(1)CLPFP has been proposed.Firstly we modeled the LEO satellite constellation.Secondly,we deeply analyzed the advantages and disadvantages and complementarities of the two load balancing strategies based on link state interaction and historical link information prediction.Combining the advantages of the two ideas,we proposed the corresponding congestion notification mechanism and link priority measurement mechanism.CLPFP designed a priority measurement mechanism to guide routing based on the dual standard of the link information of the current satellite and the congestion notification information brought by the adjacent satellite.The mechanism could select the candidate next-hop satellite with relatively light load more accurately and respond to the congestion quickly.(2)SDLBR has been proposed.Aiming at the defects of the existing centralized and distributed load-balancing routing schemes,a flexible,efficient,time-efficient and strong distributing-ability SDLBR algorithm is proposed.It mainly includes three parts:1)Distributed load awareness mechanism.This mechanism mainly realizes the network load awareness by enriching the state change packets to realize the congestion notification in a wider range,which is the premise of SDLBR algorithm.2)Two-hop routing priority measurement mechanism.This mechanism,which is the core step of SDLBR algorithm,achieves two-hop routing calculation by reconstructing the priority measurement mechanism to improve the shunt range and reduce the local optimal problem.3)Semi-distributed forwarding mechanism.This mechanism mainly realizes hop-by-hop forwarding of hop-separated computing packets by adding 3bit2 tuple data to the header,which is the fundamental guarantee of SDLBR algorithm.(3)The proposed CLPFP and SDLBR algorithms are modeled and simulated.The proposed algorithm is modeled at the network layer,node layer and process layer by using OPNET software,and the communication mechanism of iridium-like satellite network in the satellite routing stage is simulated.The simulation results are compared with the DRA algorithm?TLR algorithm and TP-MALBQR algorithm.The results show that the proposed algorithm has obvious advantages in end-to-end delay,packet loss rate,throughput,delay jitter and load distribution index.