为了应对容迟网络中拓扑结构剧烈变化、节点间连接频繁中断等问题,报文通常采用"存储—携带—转发"的方式进行传输:节点将报文存储在缓存中,携带报文直到遇到合适的机会才将报文转发给其他节点.因为缓存有限,这样的传输方式会使节点缓存溢出,导致拥塞的发生.在容迟网络环境下提出一种基于生命游戏的拥塞控制策略(game of life based congestion control strategy in delay tolerant networks,GLCCS),并将其应用于Epidemic路由方式.GLCCS借鉴生命游戏的演化思想,依据邻居节点中持有特定报文的节点比例来决定节点本地缓存中相应报文的操作.同时还提出了基于全网信息的报文排队机制和丢弃策略,依据传递或者丢弃一个报文对整个网络投递成功率的影响,计算出报文的效用值,按照效用值对缓存中报文进行排队和丢弃.在机会网络模拟器ONE中对仿真移动模型和真实运动轨迹进行模拟,实验结果表明,GLCCS与其他拥塞控制策略相比提高了投递成功率,减小了网络时延、丢包率以及负载比率.
Recently,ultrasonic waves had been introduced as the transmission medium in Body Area Networks(BANs) to reduce the incalculable damage caused by radio waves. However,the communications based on ultrasonic waves suffer from poor propagation of signals in air and consume too much energy. To address these limitations,firstly,we make the theoretical analysis to ensure ultrasonic waves could be used in BANs(UBANs). Then,we propose an error control strategy in UBANs to dynamically adjust the error control scheme and the Max-Retries based on the current channel state,which is called UECS. The UECS is based on IEEE 802.15.6 standards and considering the characteristics of ultrasonic waves in BANs. Simulation results show that UECS achieves better performance in terms of packet delivery ratio and energy consumption compared with the traditional strategies.
提出的基于马尔可夫相遇时间间隔预测的拥塞控制策略(Congestion control strategy based on Markov meeting time span prediction model,CCSMP,主要是通过规定节点缓存的排队方式和丢弃机制,将预测得到的较早与目的节点相遇的报文排于队首,尽可能丢弃效用值较低的报文,进而解决由于节点缓存有限而带来的拥塞问题。通过在ONE环境下进行仿真,与Drop-Front(DF)和Drop-Oldest(DO)两种拥塞控制策略对比表明:文中提出的拥塞控制策略提高了报文投递率,减小了平均网络时延,并且在一定程度上减少了网络负载比率和丢包率。
