东南大学学报(自然科学版)
東南大學學報(自然科學版)
동남대학학보(자연과학판)
JOURNAL OF SOUTHEAST UNIVERSITY
2015年
3期
428-432
,共5页
姚文强%杨哲%朱艳琴%李领治
姚文彊%楊哲%硃豔琴%李領治
요문강%양철%주염금%리령치
Ad hoc网络%消息传播过程%节点覆盖%高阶模型
Ad hoc網絡%消息傳播過程%節點覆蓋%高階模型
Ad hoc망락%소식전파과정%절점복개%고계모형
Ad hoc network%message propagation process%node coverage%high-order model
为避免采用随机抽样模型对Ad hoc网络消息传播过程进行定量分析时存在的高估消息覆盖节点数的问题,对现有模型进行修正,提出了一种高阶描述模型。该模型在转发消息时考虑了会对下一次消息转发到达新节点的概率产生影响的2个参数:节点被访问次数和节点度。节点被访问的次数越多,模型的阶数越高。在包含2000个节点的随机图网络拓扑上的仿真试验结果表明,当消息覆盖的节点数超过1800时,无论拓扑是否连通,一阶模型和二阶模型对Ad hoc网络消息传播过程的拟合度均优于随机抽样模型。当网络拓扑不连通时,一阶模型和二阶模型仍会高估消息覆盖的节点数;相反,当网络拓扑连通时,两者对仿真结果的拟合度较好,最终误差分别为-7%和-1%。
為避免採用隨機抽樣模型對Ad hoc網絡消息傳播過程進行定量分析時存在的高估消息覆蓋節點數的問題,對現有模型進行脩正,提齣瞭一種高階描述模型。該模型在轉髮消息時攷慮瞭會對下一次消息轉髮到達新節點的概率產生影響的2箇參數:節點被訪問次數和節點度。節點被訪問的次數越多,模型的階數越高。在包含2000箇節點的隨機圖網絡拓撲上的倣真試驗結果錶明,噹消息覆蓋的節點數超過1800時,無論拓撲是否連通,一階模型和二階模型對Ad hoc網絡消息傳播過程的擬閤度均優于隨機抽樣模型。噹網絡拓撲不連通時,一階模型和二階模型仍會高估消息覆蓋的節點數;相反,噹網絡拓撲連通時,兩者對倣真結果的擬閤度較好,最終誤差分彆為-7%和-1%。
위피면채용수궤추양모형대Ad hoc망락소식전파과정진행정량분석시존재적고고소식복개절점수적문제,대현유모형진행수정,제출료일충고계묘술모형。해모형재전발소식시고필료회대하일차소식전발도체신절점적개솔산생영향적2개삼수:절점피방문차수화절점도。절점피방문적차수월다,모형적계수월고。재포함2000개절점적수궤도망락탁복상적방진시험결과표명,당소식복개적절점수초과1800시,무론탁복시부련통,일계모형화이계모형대Ad hoc망락소식전파과정적의합도균우우수궤추양모형。당망락탁복불련통시,일계모형화이계모형잉회고고소식복개적절점수;상반,당망락탁복련통시,량자대방진결과적의합도교호,최종오차분별위-7%화-1%。
To avoid the overestimation problem of the number of nodes covered during the quantita-tive analysis of message propagation process in Ad hoc network by using the random sampling mod-el, a high-order description model is proposed by modification.In the proposed model, when the message is transmitted, the visit times and the degrees of nodes which influence the probability of the next messages transmitted to new nodes are considered.The more the visit times, the higher the or-der of the description model.The simulation results on a random graph topology network with 2 000 nodes show that when the number of nodes covered is more than 1 800 , the fitting abilities of both the 1-order model and the 2-order model for the message propagation process in Ad hoc network are better than that of the random sampling model whether the network topology is connected or not. When the network topology is disconnected, both the 1-order model and the 2-order model still over-estimate the number nodes covered; on the contrary, when the network topology is connected, the fitting degrees of both the 1-order model and the 2-order model for the simulation results are satisfac-tory, and the final fitting errors are -7%and -1%, respectively.