基于时间同步感知的智慧园区多态路由优化方法

Polymorphic Routing Optimization Method for Smart Park Based on Time Synchronization Aware

随着大量分布式能源终端接入,智慧园区中爆炸式增长的业务对队列积压、误码率、吞吐量等服务质量(quality of service,QoS)需求差异性大,现有的路由优化与差异化业务需求适配性差、时间同步与路由优化存在耦合性、链路与网络拥塞信息的不确定性等挑战,影响园区网络路由性能。针对上述挑战,提出了一种联合背压和Q学习的时间同步感知多态路由协议(backprEssure and Q-Learning based timE synChronizaTion-aware polymoRphic routIng protoCol,ELECTRIC)算法。首先,建立时间同步感知多态路由协议架构,实现园区终端的集中管理和数据传输优化;其次,根据背压算法,将队列积压最小化问题转化为队列积压差最大化问题,避免网络拥塞;然后,利用时间偏差阈值的概念对时间同步进行感知,使延迟唤醒的休眠态终端变为活跃态,并基于本地和历史信息进一步学习最优路由选择策略;最后,通过仿真结果验证,相较于基于Q学习的路由选择(Q-learning based route selection,QLRS)算法和基于队列积压的背压路由选择(queue backlog based backpressure route selection,QBBRS)算法,所提算法可分别提高平均吞吐量17.39%和56.52%,降低平均队列积压33.86%和44.07%,降低误码率31.58%和58.06%,保障智慧园区不同业务差异化的QoS需求。

With the access of a large number of distributed energy devices,the explosive growth of services in the smart park has large differences in the QoS requirements such as queue backlog,BER,and throughput.The existing smart park route selection optimization methods face challenges such as poor adaptation to the differentiated QoS requirements of services,poor coupling with time synchronization,which seriously threaten the safe and stable operation of the smart park.Motivated by the aforementioned challenges,we proposed ELECTRIC algorithm to meet the QoS requirements.Firstly,we established a time synchronization-aware polymorphic routing protocol architecture to realize centralized management of smart park devices and optimize data transmission.Secondly,according to the backpressure algorithm,we transformed the problem of minimizing queue backlog into a problem of maximizing queue backlog difference to avoid network congestion.Then,we used the concept of time error threshold to sense time synchronization,so that the dormant terminal that wakes up late became active,and further learned the optimal routing strategy based on local and historical information.Finally,simulation results demonstrate that compared with the QLRS and QBBRS algorithm,ELECTRIC can increase the average throughput by 17.39%and 56.52%,reduce the average queue backlog by 33.86%and 44.07%,and reduce the BER by 31.58%and 58.06%,which ensures the differentiated QoS requirements of different services in smart park.