为解决基于软件定义网络(software defined network,SDN)的工业物联网组播传输的可靠性问题,研究基于缓存思想的数据重传机制,提出了一种基于缓存的可靠组播方案.设计了基于缓存节点的可靠组播(cache-based reliable multicast,CBRM)框架和相应的可靠组播算法,在SDN中选择合适位置设置缓存节点并由其暂存组播数据,当接收端或中间转发设备发现数据丢失时,通过向上游缓存节点申请丢包重传以保证组播数据的可靠传输.为支持缓存节点数据重传,设计了基于UDP(user datagram protocol)的可靠传输协议URTP(UDP-based reliable transmission protocol).在Mininet仿真环境下进行了验证,验证结果表明CBRM框架和相应的可靠组播算法能在远低于传统重传成本的条件下保证组播传输的可靠性.
Abstract
To solve the reliability problem of multicast transmission in industrial Internet of things (IIoT) based on SDN (software defined network),the data retransmission mechanism based on cache is studied,and a cache-based reliable multicast scheme is proposed. A cache-based reliable multicast (CBRM) framework and the corresponding reliable multicast algorithm are designed. Cache nodes for temporarily storing multicast data are deployed in selected locations in SDN. If receiving or forwarding node detects packet loss,it will require its upstream nodes to retransmit the lost packet. To support data retransmission of cache nodes,a UDP (user datagram protocol)-based reliable transmission protocol (URTP) is designed. The results of Mininet simulation show that the CBRM framework and the corresponding reliable multicast algorithm can guarantee the reliability of multicast transmission at a much lower cost than traditional methods.
在传统组播服务中,传输过程中存在数据包重复、丢失、失序等问题,这会影响组播的可靠性并导致传输效率下降[1].软件定义网络(SDN)[2]作为一种新的网络架构通过控制器对整个网络进行集中管理,可以实现拓扑、流量、QoS(quality of service)等的实时监控,可快速应对网络故障和异常情况[3].将SDN与组播相结合,可通过控制器的集中管理和控制网络中的组播流量,在优化网络资源利用的同时提供可靠组播传输 [4-5].
现有工业SDN组播研究大多专注于框架、传输时延和可靠性.文献[6]采用多个一对一的TCP(transmission control protocol)连接实现对所有接收节点的组播通信,虽然可靠性得到了保证,但传输过程需要多次握手,且占用较多链路资源,严重影响组播数据传输时间;文献[7]提出的组播通信框架,解决了组播通信的实时性和扩展性问题,但并未考虑其可靠性;文献[8]提出利用强化学习生成最优组播树,但由于SDN的复杂性和基于机器学习方法的局限性,难以保证其准确性;文献[9]提出的基于桶的快速组播路由,提高了事件传递效率,但忽略了可靠性;文献[10]将组播树划分为多个EAM (elastic area multicast)子树,通过在子树中传输恢复数据来降低数据重传成本,但对路由成本和传输效率考虑不足,忽略了对传输时间的影响;文献[11]提出的SVC (scalable video coding)流的QoS感知组播和文献[12]提出的SDM4IIoT共享树,都实现了可扩展性,但未考虑可靠性.
为找到缓存节点保存时间和存储容量最优化对应的保存因子,利用缓存队列特性,通过NACK查找率对缓存的溢出情况进行记录.图9是不同下NACK查找率,对缓存节点类型为5的URTP数量进行统计,发现越大,缓存节点保存时间越长,缓存容量需要设置越大.当从0.7增大到1.2时,重传失败数据包数量明显降低;从1.2增大到1.4时,改善效果不明显.在权衡存储空间和丢包重传效率下,在连续发送1 000个300 B大小数据包的实验环境下,选择1.2作为值,缓存容量为20 kB.
KreutzD, RamosF M V, VerissimoP E,et al. Software-defined networking:a comprehensive survey[J]. Proceedings of the IEEE,2014,103(1): 14-76.
[4]
ZouJ F, ShouG C, GuoZ G,et al. Design and implementation of secure multicast based on SDN[C]//The 5th IEEE International Conference on Broadband Network & Multimedia Technology. Guilin,2013: 124-128.
[5]
ChitiF, FantacciR, PierucciL. Energy efficient communications for reliable IoT multicast 5G/satellite services[J]. Future Internet,2019,11(8): 164.
[6]
MahajanK, SharmaD, MannV. Athena: reliable multicast for group communication in SDN-based data centers[C]//The 9th International Conference on Communication Systems and Networks (COMSNETS). Bengaluru,2017: 174-181.
Xu Jiu-qiang, Lu Jia-xi, Li He-qun,et al. SDN-based multicast communication framework for industrial Internet of things[J]. Journal of Northeastern University(Natural Science),2023,44(2): 192-198.
[9]
ChaeJ H, KimN, LeeB D. The analysis of multicast tree construction scheme using reinforcement learning in SDN[J]. The Journal of Korean Institute of Communications and Information Sciences,2020,45(5): 820-827.
[10]
ShiY L, WongJ, JacobsenH A,et al. Topic-oriented bucket-based fast multicast routing in SDN-like publish/subscribe middleware[J]. IEEE Access,2020,8:89741-89756.
[11]
ZhangX C, YangM H, WangL,et al. An OpenFlow-enabled elastic loss recovery solution for reliable multicast[J]. IEEE Systems Journal,2016,12(2): 1945-1956.
[12]
CuiJ, KongL B, ZhongH,et al. Scalable QoS-aware multicast for SVC streams in software-defined networks[C]//2021 IEEE Symposium on Computers and Communications (ISCC). Athens,2021: 1-7.
[13]
LiH Q, LuJ X, WangJ F,et al. SDM4IIoT: an SDN-based multicast algorithm for Industrial Internet of Things[J]. IEICE Transactions on Communications,2022,105(5): 545-556.
[14]
LiaoL X, LeungV C M. LLDP based link latency monitoring in software defined networks[C]//The 12th International Conference on Network and Service Management(CNSM). Montreal,2016: 330-335.