基于动态配置等价多路径技术的无线传感器网络负载均衡算法研究

Dynamic Configure Equal Cost Multi-Paths to Achieve Load Balance in Wireless Sensor Networks

随着物联网应用的推广,作为底层核心构件的传感器网络所承载传输业务成激增趋势,使得窄带宽无线信道成为了制约物联网应用的首要因素。合理分流流量,实现负载均衡是提升网络承载能力的有效方法。本文将ECMP(Equal-Cost Multipaths)技术与传感器网络自组织特性相融合,传感器网络多跳自组织特性为业务传输提供多条等价最短路径,ECMP将业务均衡分担到这些等价最短路径上,实现负载均衡。理论证明传统ECMP配置方法全网节点开通ECMP功能不仅会增加网络控制信息开销,而且在某些情况下反而会增大区域负载,形成网络拥塞。因此,本文提出动态选择开通ECMP算法(DC-ECMP)。算法以流入节点流量等于流出节点流量作为业务守恒约束,链路带宽上限作为链路容量约束,以最大链路利用率最小化为目标函数,建立多约束优化模型。并依据最大链路使用率和节点度判定开通ECMP优先级,动态选择需开通节点,以获取最优网络传输性能。仿真结果表明DC-ECMP算法比已有PPV算法有效降低最大链路使用率,消除网络局部拥塞隐患,并且最大减少传输延时9.9 ms,节省网络资源消耗4.06%。

With the development of Internet of Things （IoTs）, wireless sensor networks, as the infrastructure of IoTs, should bear more and more various transmission services. The narrow wireless bandwidth becomes the first restrictive factors. Shunting flow to achieve load balance is the effective way to improve the bearing capacity of communication network. This paper integrated Equal-Cost Multi-Paths （ECMP）technique and WSN＇s self-organized characteristics,in which the self-organized connection and multi-hops transmission model of WSN provide more than one shortest paths from traffic source to the destination sensor node, and then ECMP can equally apportioned traffic on these available e- qual cost multi-paths to stabilize huge traffic. It proved that the traditional ECMP model, configured all of nodes with ECMP function,will increase the overhead expenses, and even make heavier traffic load in some special cases. To solve these problems,this paper proposed the dynamic configuration ECMP algorithm-DC-ECMP. In the algorithm, it defined the traffic volume conservation constraint and the wireless transmitting bandwidth upper bound constraint, made the minimizing the maximum link utilization rate as objective function, and establish the multi-constraint optimi- zation model. Using the maximum link utilization rate and node＇s degree to calculate the priority of configuration EC- MP function,each node can be dynamic configuration and achieve the optimal network＇s transmitting performance. In our evaluation with simulations,the performance of DC-ECMP is measured and compared with PPV algorithms. Based on the experimental results, DC-ECMP outperforms existing algorithms in reduce the maximum link utilization rate, short the transmitting latency 9.9 ms,and save the networks＇ resource consumption 4.06%.